Fluid Antenna Systems

• Fluid Antenna Systems are reconfigurable antennas that dynamically switch among multiple ports to harvest spatial diversity within a compact aperture.
• They implement rapid switching via mechanical, liquid-metal, or electronic techniques, enabling interference resilience and improved outage probability.
• FAS offers enhanced capacity and reliability for diverse applications, while presenting challenges in hardware miniaturization, channel estimation, and optimal port deployment.

A fluid antenna system (FAS) is an antenna paradigm in which a single radiating element can rapidly switch or “flow” among multiple spatial positions—called “ports”—over a finite physical aperture, typically much smaller than conventional antenna arrays. Unlike classic multi-antenna systems that require physically separated, fixed-position elements, FAS attains diversity gains through spatial reconfigurability, implemented either via physical movement (mechanical actuation, liquid metal, or microfluidics) or swift electronic switching among virtual or real positions. This foundation establishes FAS as a candidate physical-layer enhancement for ultra-compact, spatially adaptive, and interference-resilient wireless receivers and transmitters, especially in constrained environments or highly correlated massive connectivity scenarios.

1. Operational Principle and Mathematical Framework

FAS departs from the canonical spatial diversity paradigm by using a single radiating unit with the ability to connect to any of $N$ ports spaced across a length $W\lambda$, $\lambda$ being the carrier wavelength. The instantaneous received channel is defined as

$$|g_{\text{FAS}}| = \max\{\,|g_1|,\; |g_2|,\; ...,\; |g_N|\,\},$$

where $|g_k|$ is the random channel envelope at the $k$th port. The key performance indicator is typically the outage probability—that the post-selection SNR falls below a target $\gamma_{\text{th}}$:

$$p_{\text{out}}(\gamma_{\text{th}}) = \mathbb{P}\bigl(|g_{\text{FAS}}|^2\Theta < \gamma_{\text{th}}\bigr),$$

where $\Theta = \frac{\mathbb{E}[|x|^2]}{\sigma_\eta^2}$ is the normalized average SNR per port.

Channel correlation between ports is characterized with

$\mu_k = J_0\left(\frac{2\pi(k-1)}{N-1} W\right),$

with $J_0(\cdot)$ the zeroth-order Bessel function. The joint distribution for the port amplitudes under correlated Rayleigh fading is intricate; the paper (Wong et al., 2020) provides exact, single-integral representations for $p_{\text{out}}$ using the Marcum-Q function $Q_1(\cdot,\cdot)$, as well as practical approximate and upper-bounding expressions linking correlation and diversity.

2. Diversity and Outage Characterization

In FAS, diversity gain arises not from multiple RF chains but from spatial sampling—port selection—within a compact region, exploiting small-scale channel variation across that region. The outage probability for $N$ ports, under arbitrary spatial correlation, decays according to

$$p_{\text{out}}(\gamma_{\text{th}}) \approx \frac{1}{\det(\mathbf{J})} \varOmega^{2N},$$

in the high-SNR limit, where $\mathbf{J}$ is the spatial correlation matrix and $\varOmega$ is an SNR-dependent threshold (New et al., 2022). However, the effective diversity order is limited by the numerical rank of $\mathbf{J}$; hence, for strongly correlated ports (e.g., closely spaced), the diversity gain saturates and may be strictly less than $N$. In practice, performance can exceed that of conventional MRC ($L$-antenna maximum ratio combining) when $N$ is sufficiently large, even in a limited aperture, as confirmed by closed-form comparisons (Wong et al., 2020).

3. Capacity, Temporal Fading Statistics, and Scaling Laws

FAS manipulates the distribution of the maximal channel amplitude, affecting ergodic capacity and temporal fading metrics. The ergodic capacity is given by

$$C_{\text{FAS}} = \int_0^\infty \frac{1}{1 + y}\left[1 - \int_0^{y/\Gamma} e^{-t}\prod_{k=2}^N [1 - Q_1(\cdot,\cdot)]\,dt\right]dy,$$

where $\Gamma$ is average SNR per port and $Q_1$ expressions reflect port correlation (Wong et al., 2020). The paper also provides a closed-form lower bound.

Temporal statistics such as the level crossing rate (LCR) and average fade duration (AFD) for the FAS’s maximum-envelope process determine the rapidity of fading and expected fade times:

$$\text{LCR} = (2\sqrt{\pi} v/(\sigma \lambda)) r \exp(-r^2/\sigma^2), \qquad \text{AFD} = \frac{\mathbb{P}(r_{\text{FAS}}\leq r)}{\text{LCR}}$$

(Wong et al., 2020). These metrics, when applied to the envelope of the selected port, demonstrate that fade events in FAS—once sufficient ports are present—are both rarer and briefer, a critical property for high-reliability communications.

4. Spatial Correlation, Modeling Approaches, and Fundamental Limits

Port correlation, governed by the spatial channel structure, fundamentally constrains FAS performance. The normalized spatial correlation matrix $\mathbf{J}$’s eigenvalue spectrum determines how many quasi-independent “spatial degrees of freedom” can be harvested:

• For a normalized aperture width $W$, the effective rank (diversity order) converges to $N_{\text{eff}} = 2W + 1$ in the dense-port (oversampled) regime (Zhu et al., 10 Sep 2025).
• Increasing the port density within a fixed $W$ yields diminishing returns: after the spatial bandwidth limitation, extra ports add only redundant, highly correlated channel samples.

Analytical tractability of correlation has led to the development of block-diagonal approximations (Ramirez-Espinosa et al., 9 Jan 2024), where a large Toeplitz $\mathbf{J}$ is block-partitioned, each block yielding a dominant eigenvalue—corresponding to a spatial coherence “patch.” This enables scalable, closed-form performance analysis for both one- and two-dimensional FAS and for multiuser FAMA (Fluid Antenna Multiple Access) schemes.

5. Hardware Realizations and Implementation Issues

Physical realization of FAS is an active area of investigation. Architectures include:

• Mechanically actuated or liquid-metal-based switching, achieving continuous or pseudo-discrete spatial reconfigurability.
• Electronic or pixel-based reconfigurable antennas (PRA), as in (Zhang et al., 8 Jun 2024), where a microstrip patch feed and top pixel layer with fast RF switches realize rapid (microsecond-level) reconfiguration of radiation patterns, producing spatially correlated virtual ports with controlled covariance.
• Meta-fluid arrays (Liu et al., 15 Sep 2025) and substrate-integrated architectures capable of efficient single-RF-chain multi-user multiplexing, using programmable activation of meta-atoms and supporting ~15$\mu$s switching.

Practical constraints include finite switching speed, coupling and impedance-matching (especially in dense arrays or reconfigurable pixel structures), and physical limitations on minimum feasible port spacing. Oversampling—the deployment of more than one port per $\lambda/2$—is often required to realize full diversity; established designs must balance sampling density with switching and estimation complexity (New et al., 24 May 2024). Analytical work has further identified the existence of an optimal FAS aperture for a finite pilot/overhead budget.

6. Application Domains and System-Level Implications

FAS is emerging as a promising enabler across multiple domains:

• Single-antenna diversity enhancement and outage mitigation for ultra-compact devices.
• Multiuser scheduling in massive access/FAMA scenarios, where spatial diversity through opportunistic port selection enables interference avoidance without conventional precoding (Ramirez-Espinosa et al., 9 Jan 2024).
• Wireless-powered systems, where FAS-based port selection maximizes end-to-end product channel gain (for joint energy harvesting and data delivery) and exhibits diversity order equal to the number of ports if joint selection is used (Lai et al., 3 Feb 2024).
• Joint radar and communications (ISAC), where FAS-enabled selection and beamforming can simultaneously meet power, sensing, and communication constraints via sparse optimization frameworks, leading to quantifiable (e.g., 33%) transmit power reductions (Zou et al., 9 May 2024).
• Positioning and localization, where high inter-port correlation enables “virtual array” RSSI-processing approaches with accuracy rivalling true multi-antenna arrays (Liu et al., 2 Mar 2025).
• Error probability and coding gain analysis, where FAS achieves diversity and coding gain determined by the effective spatial rank, with tight, closed-form asymptotic SER expressions (Zhu et al., 10 Sep 2025).

7. Future Directions and Open Problems

Outstanding research challenges in the FAS domain include:

• Channel estimation and pilot overhead reduction for high-port-count FAS; approaches based on compressed sensing, AI-driven port selection, and oversampling-aware estimation are under development (New et al., 24 May 2024).
• Advanced physical and electromagnetic modeling beyond classical Bessel-function spatial correlation, accommodating non-isotropic and environment-specific effects (Ramirez-Espinosa et al., 9 Jan 2024).
• Hardware miniaturization and integration for rapid, low-power port switching and continuous reconfigurability, with emphasis on matching, isolation, and control.
• Standardization and network-layer protocol adaptation to leverage FAS spatial agility, including resource allocation, feedback, and content placement (Hong et al., 16 Jun 2025).
• Exploration of FAS integration in RIS-assisted channels, non-orthogonal multiple access, physical layer security (strategic port selection or movement for eavesdropper mitigation), and ISAC with tunable sensing/communications trade-off (Wu et al., 5 Dec 2024, Zou et al., 9 May 2024).

In summary, Fluid Antenna Systems fundamentally shift the diversity paradigm by enabling effective spatial sampling in a compact size, with performance determined by the interplay between physical aperture, port density, and spatial correlation. The carefully constructed analytical frameworks and hardware advances reviewed herein position FAS as a core technology for diversity-limited and spatially constrained future wireless systems.