Rotatable Antenna Secure Communication

• Rotatable antenna systems are dynamic arrays that physically steer beams in 2D/3D space to maximize signal quality for legitimate users while minimizing eavesdropping.
• They employ optimization techniques like alternating optimization and successive convex approximation to adjust beamforming vectors and antenna orientations effectively.
• Architectural extensions integrating RIS, metasurfaces, and six-dimensional movable antennas demonstrate enhanced secrecy rates and improved robustness in secure wireless networks.

A rotatable antenna-assisted secure communication system leverages the physical steering of antenna elements—either mechanical or electronic—to shape the spatial distribution of transmission power, thereby enhancing physical layer secrecy against eavesdroppers. By exploiting the spatial degrees of freedom provided by rotation in two or three dimensions, such systems dynamically maximize the signal-to-interference-plus-noise ratio (SINR) advantage to legitimate users and minimize information leakage towards eavesdroppers. Rotatable architectures generalize to systems with movable arrays, integrated reconfigurable intelligent surfaces (RIS), and multi-layer metasurfaces, covering both single-antenna and distributed multi-user scenarios with or without integrated-sensing requirements (Jiang et al., 22 Nov 2025, Dai et al., 14 Apr 2025, Pei et al., 18 Jan 2025, Li et al., 17 Nov 2025, Qian et al., 20 Sep 2025).

1. System Models and Fundamental Principles

Core designs model the transmitter or access point as being equipped with one or more directional antennas whose main-lobe orientation ($\theta$, azimuth and zenith) is dynamically adjustable in the three-dimensional Cartesian space. In the canonical single-user, single-eavesdropper scenario, the system places the rotatable antenna at the origin, with the legitimate user (“Bob”) and the eavesdropper (“Eve”) occupying known spatial locations $\mathbf{q}_b$, $\mathbf{q}_e$ (Jiang et al., 22 Nov 2025):

• Antenna orientation is parameterized by a one-dimensional adjustment factor, $\sigma$, dictating the boresight’s vector alignment on the line connecting $\mathbf{q}_e$ and $\mathbf{q}_b$.
• Channel models are assumed quasi-static, with Rician or Rayleigh fading. The received link SNRs at Bob and Eve, $\gamma_b(\theta)$ and $\gamma_e(\theta)$, depend on both large-scale path loss and the orientation-dependent antenna gain:

$$G_i(\theta) = \begin{cases} G_0 \cos \epsilon_i, & \epsilon_i \leq \pi/2 \ 0, & \text{otherwise} \end{cases}$$

where $\epsilon_i$ is the angle between the antenna boresight and the target direction (Jiang et al., 22 Nov 2025, Dai et al., 14 Apr 2025).

Distributed and multi-antenna extensions deploy uniform planar arrays (UPAs) with each element’s orientation ($\theta_{z,k}, \phi_{a,k}$) independently adjustable and, in advanced architectures (e.g., 6DMA), allow both the position and three-axis orientation (six DoFs) of each antenna surface to be optimized (Qian et al., 20 Sep 2025).

2. Secrecy Metrics and Rate Formulations

Secrecy performance for rotatable antenna systems is quantified via the instantaneous and average secrecy rate. In the point-to-point case: $R_s(\theta) = [C_b(\theta) - C_e(\theta)]^+,$ with $C_i(\theta) = \log_2(1 + \gamma_i(\theta))$, and [x]^+ = max{x, 0}. For practical settings where instantaneous channel state information (CSI) is not reacquired per fading block, the system fixes $\theta$ and considers the average secrecy rate over small-scale fading (Jiang et al., 22 Nov 2025).

In multi-user or colluding-eavesdropper contexts, the achievable secrecy rate generalizes to: $R_s = \left[ R_u - R_e \right]^+,$ where $R_u$ is the legitimate user’s rate and $R_e$ is the aggregate eavesdropping capacity (e.g., for the sum of eaves’ SINRs) (Dai et al., 14 Apr 2025, Qian et al., 20 Sep 2025).

Secrecy outage probability (SOP) at high SNR admits a Marcum-Q function form, enabling tight analytical bounds: $$P_\mathrm{out}(R_s) = 1 - Q_1\left( 2 \sqrt{K_b}, \sqrt{\frac{2 (1+K_b)}{E|h_b|^2} (2^{R_s}-1)} \right),$$ where $K_b$ is the Rician factor and $E|h_b|^2$ is the mean channel power (Jiang et al., 22 Nov 2025).

3. Optimization Algorithms and Theoretical Guarantees

The secrecy rate maximization problem is highly non-convex due to the coupling between beamforming vectors and rotatable antenna orientations.

• For single-antenna cases, it is proven that the objective function is quasi-concave in the one-dimensional adjustment parameter ($\sigma$), allowing global optimality via bisection search in $O(\log(\sigma_\mathrm{max}/\epsilon))$ steps (Jiang et al., 22 Nov 2025).
• For line-of-sight (LoS) only regimes, a closed-form solution for the optimal deflection angle can be derived by solving a quadratic equation that characterizes when the secrecy capacity derivative vanishes. This construction leads to three cases (exact alignment with Bob or Eve, high-SNR orthogonalization, intermediate geometries) (Jiang et al., 22 Nov 2025).
• In multi-antenna and distributed systems, secrecy maximization is decomposed into alternating optimization (AO) subproblems:
  1. Beamforming update via generalized Rayleigh-quotient optimization (principal eigenvector computation).
  2. Antenna orientation update via quadratic-constrained quadratic programming (QCQP) with successive convex approximation (SCA) for non-convex direction-vector constraints.
  3. Mode selection (in RDARS) handled with penalty-based fractional programming (FP) and SCA, employing thresholding or mixed-integer programming (Dai et al., 14 Apr 2025, Pei et al., 18 Jan 2025).
  4. Antenna placement/orientation (in 6DMA) via projected SCA and constraint linearization (Qian et al., 20 Sep 2025).

For hybrid systems integrating reflection surfaces and RIS, additional AO blocks iteratively optimize phase shifts ($\Theta$), element mode assignments, and reflection/connection configurations (Pei et al., 18 Jan 2025, Li et al., 17 Nov 2025).

4. Architectural Extensions: RDARS, Metasurfaces, and 6DMA

Generalized frameworks extend the rotatable antenna concept to:

• Reconfigurable Distributed Antenna and Reflection Surface (RDARS): Elements dynamically switch between “connection” (antenna) and “reflection” (passive RIS) modes, allowing the AO framework to jointly select element modes, beamformers, and reflection phases. Channel-aware mode selection, e.g., via closed-form thresholding, is essential for leveraging the DoF and mitigating the RIS double-fading bottleneck. Systems with only a small fraction of active (connection-mode) elements realize most secrecy gains (Pei et al., 18 Jan 2025).

• Multi-layer Transmitting RIS Integration: Rotatable arrays paired with multi-layer transmitting RIS further expand beam control capability. Secure communication is maximized under additional constraints from integrated sensing (e.g., direction-of-arrival estimation accuracy) and multi-layer phase matrix configuration. Robustness is enhanced using both model-based (Rayleigh quotient, leakage theory) and learning-based (MADDPG) optimization (Li et al., 17 Nov 2025).
• Six-Dimensional Movable Antenna (6DMA): By optimizing both 3D positions and 3D orientations of UPA surfaces, 6DMA architectures dramatically enlarge spatial beamforming DoFs. The alternating approach solves for placements and orientations (subject to physical constraints) and beamformers/artificial noise, yielding substantial SSR gains over fixed-antenna and lower-dimensional movable antenna baselines (Qian et al., 20 Sep 2025).

5. Performance Characterization and Comparative Insights

Key system-level findings across multiple configurations include:

System Type	Main Control DoFs	Secrecy Rate Gains	Complexity
Single RA (Jiang et al., 22 Nov 2025)	1D angle (σ)	Near-optimal: loss <0.01bps/Hz	O(log ε⁻¹)
RA UPA (Dai et al., 14 Apr 2025)	K×2D rotations	>Isotropic/Fixed (~>20%)	O(K^{3.5})
RDARS (Pei et al., 18 Jan 2025)	Modes/Phases/Beamformers	20–50% over RIS/DAS	Polynomial
Rotatable + RIS (Li et al., 17 Nov 2025)	Array pose, RIS, placement	~22% over fixed arrays	O(NMQ³)
6DMA (Qian et al., 20 Sep 2025)	B×(3 + 3) pos./rotations	up to 4.2 dB over FPA/others	O(NB³K²)

Higher Rician factor, higher SNR, and smaller spatial separation (angle or distance) between Bob and Eve improve achievable secrecy. Steering the boresight to “overshoot” toward Bob further degrades Eve’s reception (Jiang et al., 22 Nov 2025). In high-SNR regimes, closed-form and AO-derived solutions closely match simulation results. The robustness of the schemes extends to multi-eavesdropper and multi-user settings, with consistent relative performance over conventional architectures (Dai et al., 14 Apr 2025, Qian et al., 20 Sep 2025).

6. Design Guidelines, Implementation, and Practical Issues

• Boresight Control: Steer boresight along or “past” the Bob–Eve line to exploit directivity nulls at the eavesdropper.
• Mode/Phase Selection: Joint optimization of element modes (actively transmitting vs. passive reflecting) is critical; even partial activation offers most secrecy gains.
• Beamwidth (Directivity): Narrower beams (higher directivity factor p) improve secrecy but require fine alignment and higher hardware complexity.
• RIS/Metasurface Layers: Multi-layer RIS provides substantial gains up to 3–4 layers, after which path-loss dominates.
• CSI Requirements: Robustness to partial/inaccurate CSI is obtained via worst-case or statistical approximations in AO updates.
• Computational Complexity: All AO and bisection/hybrid methods have polynomial complexity suitable for slow-fading or static environments (update time < 100 ms for moderate array sizes).
• Practical Constraints: Mechanical rotation speed, power consumption, and angular granularity (pose quantization) are key hardware limitations. Integration of learning-based controllers (MADDPG) allows near-optimal adaptation at runtime for more complex joint configuration spaces (Li et al., 17 Nov 2025).

7. Research Directions and Extensions

Current trends in rotatable antenna-assisted secure communication span several research frontiers:

• Integrated sensing and security: Joint optimization for simultaneous wireless information and sensing (SWIS) with secrecy under strict estimation constraints (Li et al., 17 Nov 2025).
• Higher-dimensional and fully reconfigurable arrays: 6DMA opens avenues for multi-user, multi-eavesdropper, and mobile-network scenarios by leveraging maximal spatial DoFs (Qian et al., 20 Sep 2025).
• Hybrid architectures: Interfacing rotatable antennas with active and passive RIS, reconciling the trade-offs between hardware cost and beamforming flexibility (Pei et al., 18 Jan 2025, Li et al., 17 Nov 2025).
• Learning and real-time adaptation: Deep reinforcement learning frameworks are being applied to offline configuration, yielding higher secrecy rates and adaptability to non-stationary channel environments.
• Physical limits and robustness: Quantification of secrecy gains under hardware imperfections, finite-resolution actuation, and outdated/partial CSI remains ongoing (Jiang et al., 22 Nov 2025, Dai et al., 14 Apr 2025).

This body of work establishes rotatable and movable antenna architectures as a foundational approach for future physical layer secure wireless systems, providing significant advantages over fixed-antenna or passive RIS-only schemes in both static and dynamic network topologies.