Aerial Reconfigurable Intelligent Surfaces

• ARIS is a system that mounts reconfigurable intelligent surfaces on aerial platforms, enabling full-angle, 3D signal reflection for improved air-to-ground communication.
• ARIS exploits aerial mobility to overcome terrestrial limitations by enhancing LoS probability, enabling scalable aperture gains, and bypassing site acquisition constraints.
• Optimization of ARIS involves dynamic beamforming, trajectory planning, and robust channel estimation to balance signal gain against inherent path loss challenges.

Searching arXiv for recent ARIS literature to ground the article in published work. Search query: "aerial reconfigurable intelligent surface arXiv UAV RIS" Aerial reconfigurable intelligent surfaces (ARISs) are reconfigurable intelligent surfaces mounted on aerial platforms—most prominently unmanned aerial vehicles (UAVs), but also high-altitude platforms and other non-terrestrial nodes—that manipulate impinging electromagnetic waves by adjusting the amplitudes and/or phases of reflecting elements while exploiting aerial mobility for three-dimensional deployment. In the ARIS literature, the central architectural motivation is to move RIS from fixed terrestrial mounting toward mobile, elevated, and panoramic signal redirection. Compared with terrestrial RIS (TRIS), ARIS offers higher deployment flexibility, reliable air-to-ground links, and $360^\circ$ panoramic signal reflection; compared with a single UAV-borne surface, swarm-enabled aerial RIS (SARIS) addresses payload- and battery-limited aperture scaling by distributing the reflecting surface across multiple cooperating UAVs (Shang et al., 2021).

1. Architectural definition and system variants

RIS is described as a planar surface composed of many low-cost passive reflecting elements that can adjust amplitude and phase to optimize wireless signal reflection and enhance spectrum and energy efficiency. In the terrestrial case, RIS is typically fixed on facades, walls, or dedicated sites, with $180^\circ$ half-space reflection constrained by mounting geometry. ARIS instead mounts the surface on a UAV or comparable aerial platform, enabling three-dimensional positioning and full-angle reflection in the sky. The SARIS extension distributes this function over multiple UAVs, each carrying a moderate-sized RIS and cooperating as a scalable intelligent surface (Shang et al., 2021).

Configuration	Mounting	Salient properties
TRIS	Facades, walls, dedicated sites	Fixed deployment; $180^\circ$ half-space reflection
Single-platform ARIS	UAV or other aerial platform	3D placement; $360^\circ$ panoramic reflection; reliable air-to-ground links
SARIS	Multiple UAVs	Scalable aperture; cooperative communications; robustness; spatial multiplexing

This architectural progression is not merely geometric. Reviews on RIS-assisted non-terrestrial networks classify ARIS within A2G/G2A, G2G, and A2A communications, emphasizing mobile and flexible deployment, panoramic coverage, improved LoS probability, and lightweight, nearly passive implementation on aerial carriers (Ye et al., 2021). Related surveys also place ARIS within a broader 6G vision involving UAVs, HAPs, and fixed-wing aircrafts, where smart surfaces act as low-complexity signal manipulators rather than conventional RF front ends (Devoti et al., 2022, Abdalla et al., 2020).

2. Physical rationale, aperture scaling, and propagation advantages

The core physical argument for ARIS is that aerial placement reshapes the cascaded channel more favorably than fixed terrestrial deployment. Flexible and rapid positioning in 3D space can bypass site-acquisition and urban-planning constraints; elevated placement increases LoS likelihood; full-angle reflection can illuminate dead zones or obstacle-blocked regions that are inaccessible to a wall-mounted RIS (Shang et al., 2021). In non-terrestrial settings, one-reflection efficiency and panoramic geometry are repeatedly identified as decisive advantages over terrestrial surfaces that would otherwise require multi-hop or geometrically awkward reflections (Ye et al., 2021).

A principal limitation of single-platform ARIS is aperture size. Because UAVs have limited payload and battery capacity, a single carrier cannot easily support a large number of reflecting elements, so the scalability of the aperture gain is not guaranteed. SARIS addresses this by distributing the surface across $L$ UAVs, each carrying $N$ elements. In the single-user idealized regime summarized in the literature, received power scales as $N^2$ for one RIS and as $(LN)^2$ for $L$ UAVs with $N$ elements each. The same swarm architecture is also associated with high-rank MIMO channels through geometric diversity, improved resilience to failure or attack, and reduced production/control complexity relative to one massive RIS (Shang et al., 2021).

The propagation benefit does not eliminate RIS’s canonical penalty: doubled large-scale path loss, because the signal is reflected rather than regenerated. That tradeoff recurs across ARIS analyses. Aerial placement can improve geometry, LoS probability, and array gain, but the cascaded nature of the channel remains fundamental. This is why several works treat altitude, trajectory, phase design, and even element activation states as coupled optimization variables rather than independent engineering knobs (Shang et al., 2021, Aung et al., 2022).

3. Channel models, estimation burden, and quantitative performance behavior

ARIS channel modeling in the early analytical literature goes beyond purely LoS abstractions. One representative A2G model uses Nakagami-$180^\circ$0 small-scale fading and inverse-Gamma large-scale shadowing for the ground-to-air and air-to-ground hops, derives a tight approximate closed-form end-to-end outage probability, and then resorts to a DNN in mobile 3D environments where direct analysis becomes intractable. In that model, fading and shadowing strongly affect outage probability, increasing the number of reflecting elements improves energy efficiency, and increasing $180^\circ$1 from 15 to 30 reduces the required $180^\circ$2 by $180^\circ$3 dB for achieving $180^\circ$4; the ARIS-aided system also outperforms HD/FD AF/DF relaying schemes (Do et al., 2021).

SARIS introduces a distinct estimation problem because the cascaded channel dimension grows with swarm size. For a single-user, single-antenna SARIS system, the number of coefficients is $180^\circ$5; for $180^\circ$6 users it becomes $180^\circ$7, where $180^\circ$8 is the number of BS antennas, $180^\circ$9 the number of elements per UAV, and $180^\circ$0 the number of UAVs. Strong pilot contamination is highlighted as a particular difficulty because air-to-ground LoS links can cause severe pilot interference even across distant BSs. Proposed mitigations include sub-surface grouping, which reduces pilot overhead to $180^\circ$1 pilots, and sparse parametric estimation in mmWave systems based on AoA/AoD/gain structure (Shang et al., 2021).

Measured through scenario-specific metrics, ARIS performance exhibits non-monotone geometric tradeoffs. In D2D mmWave networks with human blockage, raising ARIS height reduces blockage probability exponentially, but spectral efficiency first increases and then decreases because longer paths eventually dominate. For blocker densities of $180^\circ$2, $180^\circ$3, and $180^\circ$4 bl/m$180^\circ$5, the reported optimal heights are 12, 14, and 16 m, respectively. At 1 bl/m$180^\circ$6, blockage occurrence is reduced to $180^\circ$7 from $180^\circ$8, an $180^\circ$9 reduction, and spectral efficiency reaches approximately $360^\circ$0 bps/Hz, which is $360^\circ$1 bps/Hz higher than a 5 m deployment (Nor et al., 2023). A common misconception is therefore that “higher is always better”; the literature instead treats altitude as a balance between LoS/blockage benefits and distance-dependent loss.

4. Communication, sensing, and security use cases

The canonical use cases for ARIS and SARIS are coverage extension, blockage bypass, and rapid temporary deployment. The swarm literature emphasizes cooperative panoramic full-angle reflection, multiple reflections for remote IoT, air-to-air reflection for hotspot and disaster scenarios, signal enhancement inside UAV swarms, and physical-layer security for V2V communications through adaptive aerial jamming or artificial-noise support (Shang et al., 2021). In urban backhaul, a high-altitude aerial RIS has been proposed to reflect a blocked ground-source backhaul toward multiple UAV-BSs; that system jointly optimizes aerial-RIS placement, array partitioning, and phases, with complexity upper bounded by $360^\circ$2, and reports up to $360^\circ$3–$360^\circ$4 dB transmit-power reduction over RIS-equipped terrestrial setups and more than $360^\circ$5 dB over direct terrestrial backhaul (Jeon et al., 2021).

ARIS is also used as a passive relay for status-update systems. In one IoT formulation, a UAV-mounted RIS relays updates from IoT devices to a BS, and the optimization target is expected sum Age-of-Information (AoI) through joint altitude, scheduling, and phase-shift control. That work stresses a practical architectural distinction: the RIS-UAV passive relay requires a single time slot, whereas decode-and-forward UAV relaying requires two time slots. In the absence of prior knowledge of activation patterns, PPO is used to learn altitude adjustment and communication scheduling policies, and the resulting controller outperforms random-walk and hovering-with-greedy baselines in AoI (Samir et al., 2020).

A large portion of the recent ARIS literature moves from pure communication coverage toward security and integrated non-terrestrial networking. For multi-beam satellite systems, ARIS is used to enhance secrecy by shaping the direct-plus-reflected channel of legitimate users while suppressing eavesdroppers. One line of work models channel uncertainty by moment-based ambiguity sets and reformulates distributionally robust secrecy constraints through CVaR before alternating transmit and reflective beamforming updates (Wang et al., 29 May 2026). Another considers multiple ARISs in multi-group satellite communication and jointly optimizes transmission and reflection beamforming, ARIS-group association, and ARIS deployment by block coordinate descent (Wang et al., 28 Aug 2025). Anti-jamming is addressed in large-scale ARIS swarms through a mean-field deployment model with density $360^\circ$6, an adaptive jammer with a proximity-directivity trade-off, and a robust max-min design that jointly optimizes BS beamforming, ARIS reflection, and ARIS spatial distribution; the resulting deployment follows a spatial water-filling principle and has complexity independent of the number of UAVs (Luo et al., 12 Sep 2025).

Beyond communications, aerial intelligent surfaces are increasingly tied to sensing. In stationary-radar SAR imaging, a UAV-mounted active RIS creates a virtual LoS path and a virtual synthetic aperture; a range-Doppler imaging algorithm together with FP/MM-based coefficient optimization is used to maximize SNR under maximum-power and amplification constraints (Sun et al., 2024). In low-altitude ISAC networks, UAV-mounted IRS combined with movable antennas is presented as a way to jointly optimize channel reconstruction, tracking precision, and robust communication-sensing resource allocation under mobility and clutter (Wang et al., 13 Nov 2025).

5. Optimization and control methodologies

ARIS design problems are characteristically mixed-integer, non-convex, and trajectory-coupled. The foundational SARIS treatment isolates three recurring blocks: beamforming design, channel estimation, and deployment/movement. Deployment is cast as 3D positioning that balances path loss, LoS/NLoS probability, and coverage; movement is modeled either as tightly coordinated swarm motion around a reference point or as sub-swarm operation; and reinforcement learning is recommended when trajectory and beamforming decisions are strongly sequential (Shang et al., 2021).

Subsequent work refines this methodological picture rather than replacing it. Energy-efficient multiple-ARIS downlink design decomposes joint deployment, on/off element states, phase shifts, and power control into three subproblems solved alternately by SCA, actor-critic proximal policy optimization (AC-PPO), and whale optimization algorithm (WOA). That study reports $360^\circ$7 sum-rate gain over ARIS with fixed phase shifts, $360^\circ$8 over single-ARIS systems, $360^\circ$9 energy-efficiency gain over single-ARIS, $L$0 over UAV-Relay, and $L$1 sum-rate gain over a multiple ground-RIS scenario (Aung et al., 2022). In dual-aerial active-surface ITNTN, BCD combines WMMSE for transmit beamforming, manifold-based RCG for phase shifts, SCA for amplification factors, and first-order trajectory approximations, with an average sum-rate improvement of approximately $L$2 over passive RIS under given power constraints (Saiprudhvi et al., 14 Apr 2026).

Security- and robustness-oriented ARIS designs introduce a different algorithmic toolkit. Distributionally robust satellite secrecy relies on CVaR-based reformulation, SDR, FP, and penalty-based CCP within an AO loop (Wang et al., 29 May 2026). EMF-aware uplink MU-MIMO combines Dinkelbach fractional programming for beamforming parameters, quadratic transform plus SDR for phase design, heuristic resource allocation, KKT-based power allocation, and SCA for UAV trajectory; its optimized ARIS architecture reduces EMF exposure by over $L$3 relative to fixed ARIS and over $L$4 relative to non-ARIS schemes (Chemingui et al., 2024). Large-scale anti-jamming ARIS replaces discrete UAV positions by a continuous density function $L$5 satisfying $L$6, then derives a thresholded optimal deployment $L$7 from a net marginal gain function $L$8, yielding a spatial water-filling rule (Luo et al., 12 Sep 2025).

A recurrent theme across these methods is timescale separation. Slow variables include deployment region, altitude, swarm geometry, or trajectory; fast variables include beamforming, RIS phases, and resource allocation. This suggests that scalable ARIS control will continue to combine geometry-aware large-scale planning with fast local adaptation rather than solving a monolithic real-time problem.

6. Limitations, misconceptions, and open research directions

The ARIS literature is explicit that aerial mobility does not eliminate the classical RIS bottlenecks. Payload, battery, and controller complexity constrain the number of elements on a single carrier; passive operation still incurs doubled large-scale path loss; and real-time control of trajectory plus phase configuration leads to severe CSI and signaling overhead. Surveys of aerial smart surfaces further identify onboard control overhead, variable battery drain, antenna-array size limits, beam misalignment due to aerial perturbation, and the difficulty of low-latency CSI acquisition for nearly passive surfaces (Shang et al., 2021, Devoti et al., 2022, Abdalla et al., 2020).

Several scenario studies directly correct intuitive but misleading simplifications. First, ARIS is not universally superior to terrestrial deployment in every regime: higher mobility and full-angle reflection come with more dynamic and fluctuating channels, shorter lifetime due to aerial-platform energy limits, and increased synchronization complexity (Ye et al., 2021). Second, larger altitude is not monotonically beneficial, as the D2D human-blockage analysis shows. Third, ideal diagonal cascaded-channel assumptions are increasingly regarded as insufficient. Open modeling directions explicitly include fading, correlation, polarization, incident angle, and UAV wobbling effects, together with field validation of ARIS and SARIS channel models (Shang et al., 2021).

Current research directions therefore concentrate on practical robustness rather than only asymptotic gain. The open problems repeatedly identified include robust beamforming under CSI uncertainty, scalable channel estimation with limited RF chains, air-ground integrated RIS networks that combine TRIS and SARIS, integration with edge computing, wireless power transfer, symbiotic radio, secure communications, predictive and event-driven control, and low-complexity two-timescale designs for highly dynamic ISAC and NTN settings (Shang et al., 2021, Wang et al., 13 Nov 2025). A plausible implication is that ARIS will remain less a single device class than a design space spanning passive, active, single-platform, and swarm-based aerially deployed intelligent surfaces, unified by the attempt to jointly optimize electromagnetic reflection and 3D network geometry.