Movable Antenna Systems (MAS)

• Movable Antenna Systems (MAS) are dynamic wireless architectures that reposition antenna elements to actively manipulate signal phases and improve channel responses.
• MAS leverages mechanical actuators and advanced optimization algorithms to enhance geometric alignment, multipath diversity, and counteract deep fading.
• MAS applications span integrated sensing and communications, IoT networks, and V2X systems, demonstrating improvements in throughput, sensitivity, and coverage.

A movable antenna system (MAS) is a wireless transceiver architecture in which each antenna element is physically repositioned within a confined region, typically on the order of a few wavelengths, by mechanical actuators such as MEMS sliders, robotic arms, or precision nano-positioners. This capability introduces real-time, controllable spatial degrees of freedom (DoFs) beyond conventional fixed arrays. By directly manipulating the spatial configuration of the array, MAS enables the reconfiguration of the channel response, augmentation of received signal strength, and circumvention of deep fades, all without additional spectrum usage or modifications to passive nodes in the environment. MAS is particularly impactful for integrated sensing and communications (ISAC), backscatter-based networks, and emerging ultra-low-power wireless IoT applications, where geometric alignment and multipath diversity are critical (Zhang et al., 27 Jan 2026).

1. Physical and Functional Principles

A MAS fundamentally differs from fixed-position antenna arrays in that each antenna’s phase center can be actively and continuously repositioned. Sub-wavelength displacements induce path length changes on the order of $d \cos\theta$ (where $d$ is the displacement and $\theta$ the angle to the incident or scattered wave), directly modulating the phase accumulation and thus the total channel response. At the system level, this mechanism enables manipulation of:

• Double fading in cascaded channels: In backscatter ISAC, MAS directly addresses the compounded fading inherent to cascaded forward-backward links, escaping deep nulls caused by geometric misalignment between transceivers and passive tags.
• Incident and receive angle tuning: MAS enables fine adjustment of the incident ($\theta_i$) and receive ($\theta^r$) angles at the tag or target, exploiting the angular dependence of radar cross-section (RCS) and maximizing integrated gains.
• Motion-enabled angular diversity: By sequentially probing spatial regions, MAS allows the system to "hunt" for spatial nulls or peaks, maximizing SNR or task-specific metrics on a per-coherence-interval basis.

This spatial DoF can be exploited systematically, for instance in a closed-loop cycle:

1. Probe candidate antenna positions and measure SNR;
2. Optimize by selecting the position maximizing a target metric (e.g., SNR, communication rate, sensing CRB);
3. Move the physical elements to the optimal configuration and repeat every coherence interval (Zhang et al., 27 Jan 2026).

2. Channel Modeling and Mathematical Formulation

The channel model for MAS is derived from the field-response formalism. Consider a single-antenna transmit MAS element at position $p_t(d_t)$, a passive tag at $p_{\text{tag}}$, and a receive MAS element at $p_r(d_r)$. The forward and backward links are represented as

$$h_1(d_t) = \rho_1(d_t) e^{-j2\pi \|p_t(d_t) - p_{\text{tag}}\|/\lambda}$$

$$h_2(d_r) = \rho_2(d_r) e^{-j2\pi \|p_r(d_r) - p_{\text{tag}}\|/\lambda}$$

where $$\rho_1(d_t) = \lambda/(4\pi\|p_t(d_t) - p_{\text{tag}}\|)$$ and similarly for $\rho_2$. The cascaded (multiplicative) backscatter channel is then

$h(d_t, d_r) = \alpha \cdot h_1(d_t) \cdot h_2(d_r)$

with $\alpha$ the complex reflection coefficient of the tag load. The resulting received signal is

$y = \sqrt{P} \, h(d_t, d_r) \, x + n$

For $N_t \times N_r$ MAS arrays, array steering vectors $A_t(d_t)$, $A_r(d_r)$ generalize the model: $H(d_t, d_r) = \alpha \cdot A_r(d_r) A_t^T(d_t)$ Performance is dominated by the double fading term $$|h|^2 \propto 1 / (\|p_t - p_{\text{tag}}\|^2 \|p_r - p_{\text{tag}}\|^2)$$, so small changes in geometry induce significant SNR fluctuations—an effect MAS is engineered to exploit (Zhang et al., 27 Jan 2026).

3. Real-Time Motion Control and Optimization Techniques

MAS operation demands efficient position selection algorithms adapted to the non-convex, high-dimensional channel landscape:

• Exhaustive/Hierarchical Search: Partition the movement region into a finite grid, evaluate SNR at all candidate positions, and select the maximizer. Hierarchical versions apply multiresolution tree searches.
• Gradient/Coordinate Descent: Numerically estimate gradients of $|h|^2$ w.r.t. position, iteratively move toward local maxima; particularly effective in lower-dimensional movement spaces.
• Reinforcement Learning: Use Q-learning or policy-gradient RL, where states are current channel estimates or SNR maps and actions are steps in allowed position directions; reward functions encode communication or sensing performance.
• Joint Waveform-Motion Co-design: Alternate optimization of spatial positions, waveform design, and power-split ratios in composite objective functions; enables balancing mechanical actuation cost, RF power, and sensing accuracy (Zhang et al., 27 Jan 2026).

Computational complexity is managed via discrete sampling, graph-based reformulations, and suboptimal but fast heuristics, e.g., graph dynamic programming or linear sequential-update for one-dimensional rails (Mei et al., 2024).

4. Performance Gains and Comparative Evaluation

MAS introduces substantive performance improvements validated in urban microcell simulations:

• Throughput: MAS achieves a 20–30% increase in communication rate compared to static arrays, by evading deep double-fading nulls.
• Sensing SNR: Up to 5 dB gain in angle/range estimation, owing to geometry-adaptive aperture enlargement and spatial path recovery.
• Coverage: Extended tag detection range by approximately 20% due to dynamic SNR maximization.
• Robustness: Outperforms static configurations across a broad range of communication-sensing power split factors ($\rho \in [0.1,0.9]$), showing resilience to geometric and resource division trade-offs (Zhang et al., 27 Jan 2026).

These gains are robust to a variety of scenarios and reflect the core advantage: active geometry adaptation without the need for additional RF resources or information updates from passive tags.

5. Application Scenarios and System Architectures

MAS is critical for robust ISAC in several IoT deployment classes:

• Urban Transportation: MAS-equipped roadside units maintain reliable V2X communication with passive tags despite vehicle mobility and environmental blockages.
• Industrial Logistics: MAS-based readers slide along warehouse racks, overcoming metal obstructions and enabling continuous inventory tracking and precise localization.
• Human-Centric Sensing: MAS ensures robust communication and sensing with wearable tags, dynamically compensating for occlusion and movement.
• Low-Altitude and Aerial IoT: MAS on drones or ground stations maintain stable links under spatial dynamics, enabling airspace surveillance and drone traffic management (Zhang et al., 27 Jan 2026).

The generic MAS-assisted B-ISAC architecture comprises a MAS-equipped full-duplex AP/reader and MAS receiver, serving passive tags by coordinated spatial probing, adaptive repositioning, and closed-loop processing for joint communication and geometric sensing.

6. Open Research Problems and Future Directions

Emerging research vectors for MAS include:

• Motion-aware channel modeling and digital-twin environments: Predicting channel evolution under MAS motion for proactive adaptation.
• Integrated motion, waveform, and resource optimization: Joint frameworks trading off actuation delay/energy, RF power, and sensing latency in real time.
• Scalable, distributed MAS control: Multi-tag, multi-transceiver coordination to harness spatial flexibility system-wide.
• Robust hardware/software design: Waveforms and receivers that maintain performance under mechanical jitter, integrating backward-compatibility with fixed ISAC and IRS infrastructure.
• Security and privacy enhancement: Leveraging spatial agility for secure key extraction, jamming avoidance, and protecting movement pattern inference in adversarial environments (Zhang et al., 27 Jan 2026).

MAS thereby transforms the geometric bottlenecks of backscatter-based and ISAC links into an actively controllable design dimension, unlocking new avenues in spatially adaptive, robust, and energy-efficient wireless networks for next-generation low-power IoT.