Rotatable Antennas (RAs): Principles & Applications

• Rotatable Antennas (RAs) are antennas with dynamically adjustable boresight that enhance directivity and flexible spatial control.
• They integrate both mechanical and electronic methods, such as servo motors and tunable materials, to optimize beamforming and extend coverage.
• RA systems demonstrate significant gains in spectral efficiency, interference suppression, and sensing resolution across diverse wireless applications.

A rotatable antenna (RA) is an antenna or antenna array whose physical orientation—specifically, the boresight direction—can be dynamically adjusted through mechanical or electronic means. In contrast to fixed-orientation antennas, RAs offer additional spatial degrees of freedom (DoF) by providing physical orientation control of either the entire array or individual elements. This property enables dynamic manipulation of the array’s directivity pattern, yielding substantial improvements in communication and sensing by enhancing desired links, suppressing interference, extending coverage, and increasing spatial multiplexing and sensing resolution. RAs have been investigated across diverse system models, including conventional MIMO, cell-free networks, ISAC, cognitive radio, and secure communication, with both theoretical and experimental demonstrations showing quantifiable performance gains over fixed-arrays (Tominaga et al., 2024, Zheng et al., 5 Jan 2025, Zheng et al., 22 May 2025, Zhang et al., 5 Sep 2025, Wen et al., 13 Dec 2025, Peng et al., 23 Jan 2026, Zhou et al., 13 Mar 2025, Dai et al., 28 Feb 2025, Dai et al., 24 Feb 2025).

1. Rotatable Antenna Principles and Hardware Architectures

Rotatable antennas generalize conventional antenna systems by adding explicit control over the physical orientation of either each element or the whole array. Mechanically, this typically involves servo-motor platforms, gimbals, or MEMS actuators that allow rotation in one or more axes (azimuth, elevation, or both) (Zheng et al., 22 May 2025, Dai et al., 24 Feb 2025). Electronically, multi-feed structures, tunable materials (e.g., liquid crystals), or electronically tuned parasitic arrays offer fast but limited steering range (Zheng et al., 22 May 2025). Hybrid schemes exploit both mechanical and electronic control for wide-angle, low-latency steering.

Beyond element-level actuation, array-level rotation (where the entire array rotates as a rigid body) offers system-level DoF with low mechanical complexity but less flexibility. Cross-linked RA architectures decouple azimuth and elevation control at the element or panel level to greatly reduce hardware complexity, achieving performance close to fully flexible designs with hardware DoF scaling only as $M+N$ for an $M\times N$ array (Zheng et al., 8 Jan 2026).

Key architectural types:

Architecture	Orientation DoF	Mechanism
Element-level rotation	$2N$ (az, el per elem)	Servo, MEMS, etc.
Array-level rotation	2 (az, el global)	Single actuator
Cross-linked (CL)	$M+N$ (row & col)	Shared actuators
Mixed/Hybrid	Macro (global) + Micro (element)	Co-design

2. Mathematical Models and System Integration

RA-based systems are modeled by incorporating the orientation-dependent array manifold. In 3D, each RA axis is parameterized by azimuth ($\varphi$) and zenith/elevation ($\theta$), with unit vector $$\mathbf{f}(\theta, \varphi) = [\sin\theta\cos\varphi, \sin\theta\sin\varphi, \cos\theta]^T$$. The directional element gain for an offset angle $\varepsilon$ is typically modeled as $G(\varepsilon) = G_0 \cos^{2p}(\varepsilon)$ for $\varepsilon \in [0, \pi/2)$, with $G_0=2(2p+1)$ enforcing power conservation and $p$ the directivity factor (Zheng et al., 5 Jan 2025, Zheng et al., 22 May 2025).

Channel coefficients for user $k$ and element $n$ include path-loss, small-scale fading, and orientation-dependent gain:

$$h_{k,n}(\mathbf{f}_n) = \sqrt{\beta_{k,n} G(\varepsilon_{k,n})} e^{-j2\pi r_{k,n}/\lambda}$$

with $\beta_{k,n}$ denoting path loss and $$\cos\varepsilon_{k,n} = \mathbf{f}_n^T \mathbf{u}_{k,n}$$.

System models have been constructed for:

• Uplink and downlink MIMO with per-element or global rotation (Wen et al., 13 Dec 2025, Tominaga et al., 2024)
• Cell-free networks: distributed APs each with rotatable elements, exploiting macro-diversity (Peng et al., 23 Jan 2026, Pan et al., 4 Dec 2025)
• Cognitive radio: joint orientation and beamforming to maintain secondary-user SINR under interference constraints (Tan et al., 30 Sep 2025, Peng et al., 24 Sep 2025)
• ISAC: multi-objective optimization to simultaneously maximize sum-rate and minimize the CRB for angle estimation (Zhou et al., 13 Mar 2025, Wang et al., 10 Sep 2025)

3. Optimization and Algorithmic Strategies

RA-induced spatial DoFs give rise to nonconvex optimization problems for control of both orientation vectors $\{\mathbf{f}_n\}$ and digital beamformers $\{\mathbf{w}_k\}$, typically subject to power, mechanical, and practical constraints. Alternating optimization (AO) is the dominant technique, cycling between:

• Beamformer update (MMSE, ZF, WMMSE, FP) for fixed orientations
• Orientation update (gradient ascent, Frank-Wolfe on the unit sphere/spherical cap, SCA, or global search methods)

Closed-form updates exist in some special cases (e.g., single-user MRC aligns all RA boresights with the LoS direction), while general cases use iterative convex surrogates or particle swarm optimization (PSO) (Tominaga et al., 2024, Zheng et al., 5 Jan 2025, Zhou et al., 13 Mar 2025, Zhang et al., 5 Sep 2025).

For joint beamforming and orientation with discrete-resolution actuators, cross-entropy optimization (CEM) and codebook-based schemes are effective, outperforming simple nearest-projection to grid (Peng et al., 24 Sep 2025). In hybrid beamforming (sub-connected analog/digital), FP-based AO with gradient ascent over rotation angles delivers near-optimal performance (Wang et al., 10 Sep 2025).

4. Impact on Wireless Communication and Sensing

Extensive simulation and experimental work demonstrates significant performance gains attributable to RA-enabled systems:

• Spectral Efficiency: In Rician fading, array rotation recovers 30–40% higher mean SE over fixed arrays for moderate/large K-factor, and outperforms 2D MAAs when movement area is restricted (Tominaga et al., 2024).
• Interference Management: RA-driven null steering via 3D orientation can maintain high main-lobe gain towards desired users while nulling multiple interferers, vastly relaxing angular separation constraints (Wen et al., 13 Dec 2025).
• Sum-Rate and Fairness: In cell-free MIMO, jointly optimized RA orientation and beamforming yield worst-user rate improvements of 24–53% over fixed-directional or isotropic baselines, with further logarithmic utility fairness improvements via low-complexity Frank-Wolfe AO (Peng et al., 23 Jan 2026).
• Integrated Sensing and Communication (ISAC): RA rotation enables CRB minimization by increasing effective spatial aperture, yielding 15–20% higher sum-rate and up to 40% reduction in localization CRB over beamforming-only setups (Zhou et al., 13 Mar 2025).
• Spectrum Sharing and Secrecy: Joint design of RA orientation and transmit vectors enables strict interference temperature capping at primary users alongside enhanced link SINR for secondary users, with secrecy rate and spectrum sharing gains scaling with array size and directivity (Tan et al., 30 Sep 2025, Dai et al., 14 Apr 2025).
• Physical Layer Prototyping: Radar-aided and vision-guided RA prototypes deliver 7–15 dB SNR gains and up to 3x angular coverage extension in practical indoor environments (Dai et al., 28 Feb 2025, Dai et al., 24 Feb 2025).

5. Experimental Platforms and Prototyping

Real-system demonstrations validate theoretical predictions. Prototypes combine directional patch antennas mounted on pan-tilt gimbals, controlled by MCU platforms and closed-loop servo systems. Angle-of-arrival is acquired via visual recognition (DeepSORT/YOLO) or TOF LiDAR, guiding rapid orientation adjustments. Measured SNR gains versus fixed are 7–15 dB at large off-boresight angles; coverage angle is expanded from 60° to 180° for practical indoor scenarios (Dai et al., 28 Feb 2025, Dai et al., 24 Feb 2025). Servo and processing latency are on the order of tens of milliseconds, confirming feasibility for dynamic wireless environments.

6. Design Trade-offs, Constraints, and Guidelines

Key system design issues include:

• Orientation Range: Most RA-induced gains are realized with modest zenith or azimuth ranges ($\theta_{\max}\approx\pi/10$ to $2\pi/10$), maintaining high performance at low mechanical cost (Peng et al., 23 Jan 2026, Zheng et al., 5 Jan 2025).
• Directivity vs. Control Resolution: High-gain, narrow-beam RAs require finer orientation control to avoid performance degradation from misalignment (Dai et al., 14 Apr 2025).
• Hardware Complexity: Cross-linked and array-level rotation architectures offer favorable trade-offs, retaining most of the DoF at much lower actuation count (Zheng et al., 8 Jan 2026).
• Channel Estimation: RAs enable enhanced environmental probing during channel training, facilitating more accurate CSI and angular resolution when their orientations are adaptively cycled (Xiong et al., 25 Jun 2025).
• Control Algorithms: Continuous-time, fine-grained steering can be traded off against discrete codebooks and predictive or learning-based control for different deployment scenarios (Peng et al., 24 Sep 2025, Zheng et al., 22 May 2025).

7. Applications and Future Outlook

RA technologies are positioned as foundational tools for:

• Cell-free massive MIMO with macro-diversity: Homogenizing link quality via orientation-aware distributed beamforming (Peng et al., 23 Jan 2026).
• ISAC systems: Supporting stringent sensing and communication requirements by dynamically balancing channel and localization accuracy (Zhou et al., 13 Mar 2025, Wang et al., 10 Sep 2025).
• 5G/6G spectrum sharing: RA-driven spatial resource management for cognitive radio and multi-user MISO interference channels (Tan et al., 30 Sep 2025, Peng et al., 24 Sep 2025).
• Adaptive coverage in low-altitude and UAV-aided networks: Flexible beamforming and near-field focusing for agile 3D cellular and UAV networks (Li et al., 1 Nov 2025).

Active research topics include high-speed electronic RA materials, learning-based closed-loop orientation control, integrated approaches with reconfigurable intelligent surfaces, and Joint ISAC optimization for next-generation networks (Zheng et al., 22 May 2025). Early-stage challenges remain in the areas of optimal joint scheduling of rotation and beamforming, robust CSI acquisition under mobility, and low-latency, energy-efficient hardware integration.

References:

(Tominaga et al., 2024, Zheng et al., 5 Jan 2025, Zheng et al., 22 May 2025, Wen et al., 13 Dec 2025, Zhang et al., 5 Sep 2025, Peng et al., 23 Jan 2026, Zhou et al., 13 Mar 2025, Dai et al., 28 Feb 2025, Dai et al., 24 Feb 2025, Pan et al., 4 Dec 2025, Zheng et al., 8 Jan 2026, Dai et al., 14 Apr 2025, Xiong et al., 25 Jun 2025, Tan et al., 30 Sep 2025, Peng et al., 24 Sep 2025, Wang et al., 10 Sep 2025, Li et al., 1 Nov 2025, Wu et al., 2024).