PaperlayerView exact revision on arXiv

Source-linked AI summary

A Contemporary Survey on Fluid Antenna Systems: Fundamentals and Networking Perspectives

Hanjiang Hong, Kai-Kit Wong, Hao Xu, Xinghao Guo, Farshad Rostami Ghadi, Yu Chen, Yin Xu, Chan-Byoung Chae, Baiyang Liu, Kin-Fai Tong, Yangyang Zhang

arXiv:2506.13317v2cs.ITeess.SP

TL;DR

Next-generation networks require higher rates, lower latency, and greater connectivity, while FAS research still lacks a comprehensive networking perspective. This paper surveys FAS application scenarios, physical-layer fundamentals, multi-user FAMA, and networking techniques, concluding that FAS spans single-user HomoNets, multi-user HomoNets, and HetNets while networking solutions remain an active area of interest. Practical development is constrained by immature hardware and the scarcity of empirical channel models.

  • Problem

    FAS networking technologies remain insufficiently surveyed despite growing interest in integrating FAS into future 6G networks.

  • Method

    The paper conducts a contemporary survey covering FAS application scenarios, channel modeling and estimation, single-user configurations, FAMA, and networking techniques including QoS, power allocation, and content placement.

  • Results

    The survey categorizes FAS applications into single-user HomoNets, multi-user HomoNets, and HetNets and summarizes corresponding physical-layer and networking techniques.

  • Takeaways & Limitations

    FAS provides a broad platform for spatially reconfigurable wireless designs, while networking solutions are needed to support efficient, reliable, and secure practical operation.

  • Takeaways & Limitations

    Empirical FAS channel models remain scarce because current hardware prototypes are insufficiently mature and face durability and miniaturization issues.

Abstract

from arXiv · show

The explosive growth of teletraffic, fueled by the convergence of cyber-physical systems and data-intensive applications, such as the Internet of Things (IoT), autonomous systems, and immersive communications, demands a multidisciplinary suite of innovative solutions across the physical and network layers. Fluid antenna systems (FAS) represent a transformative advancement in antenna design, offering enhanced spatial degrees of freedom through dynamic reconfigurability. By exploiting spatial flexibility, FAS can adapt to varying channel conditions and optimize wireless performance, making it a highly promising candidate for next-generation communication networks. This paper provides a comprehensive survey of the state of the art in FAS research. We begin by examining key application scenarios in which FAS offers significant advantages. We then present the fundamental principles of FAS, covering channel measurement and modeling, single-user configurations, and the multi-user fluid antenna multiple access (FAMA) framework. Following this, we delve into key network-layer techniques such as quality-of-service (QoS) provisioning, power allocation, and content placement strategies. We conclude by identifying prevailing challenges and outlining future research directions to support the continued development of FAS in next-generation wireless networks.

I. Introduction

FAS introduces dynamically reconfigurable antenna position and structure to add spatial degrees of freedom for adapting wireless systems to changing channel conditions. The survey reviews FAS motivations, application scenarios, physical configurations, and networking-oriented research directions.

  • Motivation: 6G targets include a peak data rate of 1 Tbps, 1 ms end-to-end latency, and 10 million devices per km2, motivating new wireless technologies.The paper frames these objectives as requirements for applications involving interaction among humans, machines, and environments.
  • FAS Fundamentals: FAS dynamically reconfigures antenna position and structure to provide additional spatial degrees of freedom beyond fixed-position antennas.Its software-defined architecture can use fluidic, conductive, or dielectric materials to reconfigure antenna position and shape in real time.
  • Survey Scope: The survey addresses a literature gap on networking technologies specifically designed for FAS and examines their integration into future 6G networks.It covers FAS scenarios, fundamentals, and networking techniques intended to harness its potential.
  • Basic FAS: Basic FAS configurations include SISO-FAS with one activated port and MIMO-FAS with multiple activated ports, using port selection to enhance receiving performance.FAS ports may be arranged as one-dimensional linear or two-dimensional planar configurations.
  • Basic FAS: Early studies report that SISO-FAS outage probability decreases as the number of ports increases and that sufficiently many ports can outperform maximal-ratio combining.Subsequent work also examined capacity, level crossing rate, average fade duration, and machine-learning-assisted port selection.

2) Case 1B: RIS-aided FAS:

FAS application scenarios extend from RIS-assisted links and index modulation to multi-user MIMO, FAMA, and ISAC, using spatial reconfigurability to improve communication performance.

  • RIS-aided FAS: RIS-assisted FAS mitigates the double-path loss in cascaded RIS systems when obstacles block the direct BS–user link.The RIS redirects the RF signal to the mobile FAS receiver.
  • FAS-aided IM: FAS-aided index modulation uses selected fluid-antenna ports as information-bearing entities to enhance spectral efficiency.Later work considers multiple-port selection, MIMO, OFDM, neural classification, and RIS-assisted transmission.
  • MU-MIMO-FAS: MU-MIMO-FAS uses spatial reconfigurability at the BS, UEs, or both sides to improve robustness, reduce estimation errors, and increase uplink multiple-access capacity.Dynamic repositioning can reduce hardware imperfections, while compressed sensing supports sparse channel recovery.
  • FAMA: FAMA lets users select ports with weak interference while reusing the same frequency resources, with fast and slow variants trading responsiveness against practicality.Fast FAMA switches per symbol and can support hundreds of users, whereas slow FAMA switches once per coherence time and supports tens of users.
  • FAS-aided ISAC: FAS-aided ISAC dynamically balances sensing and communication, shifting the ISAC Pareto frontier and enabling adaptive performance tuning.Reinforcement-learning port selection achieved 30% gains on spectral efficiency in the cited study.

1) Case 3A: Content-centric FAS HetNets:

Content-centric FAS HetNets place popular content near users while using reconfigurable antennas to optimize links, spectral efficiency, reliability, and robustness. The section also reviews channel models whose analytical simplicity can trade off against accuracy.

  • Content-centric HetNets: Caching popular content at small-cell BSs or edge nodes reduces end-to-end latency and backhaul load in content-centric HetNets.Users connect to the nearest SBS containing the requested content, with content popularity modeled by a Zipf distribution.
  • Content-centric HetNets: FAS enables adaptive link establishment between SBSs and UEs by selecting antenna positions that enhance received signal strength during content retrieval.The same adaptability supports spectral efficiency and reliable communication in high-mobility settings.
  • Channel modeling: FAS channel models cover rich- and finite-scattering environments, including 1D-FAS and 2D-FAS configurations with spatially correlated ports.The reviewed models include point-to-point settings with a fixed-position transmitter and FAS receiver, as well as extensions to both-end FAS and MIMO-FAS.
  • Channel modeling: Simplified FAS channel models improve analytical tractability but can produce overly optimistic performance evaluations by overlooking inter-port dependencies.More accurate models are more complicated, while some approximations retain only a reduced set of dominant eigenvalues.
  • Channel modeling: Common-parameter correlation models establish mutual correlations among all ports, but one reviewed model still inadequately reflects the correlation structure characterized by the reference formulation.The survey notes that these models support analyses of outage probability, multiplexing gain, and port selection.
  • Channel modeling: Eigen-decomposition-based channel representations approximate correlated FAS channels using weighted combinations of independent complex Gaussian variables.The formulation uses the covariance matrix eigenvectors and eigenvalues to generate the channel coefficients.

2) Finite-Scattering Environment:

Finite-scattering FAS channel estimation exploits sparse propagation to reconstruct full-port CSI from limited measurements, while balancing accuracy, overhead, and computational cost.

  • Finite-Scattering Environment: Finite-scattering FAS channels can be modeled geometrically, with sparse path parameters enabling CSI reconstruction at all ports from preset estimating locations.The channel model uses propagation paths, complex gains, and transmitter- and receiver-side steering vectors.
  • Finite-Scattering Environment: Using K ≪ N estimating locations reduces measurement requirements in sparse environments, while LS accuracy depends on pilot repetitions and transmit power.The LS approach becomes impractical when the number of ports is large because of hardware complexity and pilot overhead.
  • Finite-Scattering Environment: L3SCR estimates path counts and AoAs, then estimates AoDs and gains before reconstructing the complete channel matrix.Its stages combine DFT-based estimation, angle rotation, matched filtering, and the planar-wave geometric model.
  • Finite-Scattering Environment: OMP improves estimation accuracy when the BS has insufficient antennas, but its matrix inversions make it substantially more computationally expensive than L3SCR.L3SCR accuracy depends on the number of BS antennas, whereas OMP trades higher complexity for improved estimation.
  • Finite-Scattering Environment: LS can achieve lower NMSE than L3SCR and OMP in many scenarios, but requires all users to switch across all N ports and transmit pilots.This hardware switching and pilot overhead can reduce spectral efficiency.

C. Single-user FAS and its Variants

Single-user FAS ranges from one-port SISO operation to multi-port MIMO operation, using port selection and joint beamforming or power allocation to improve rate.

  • SISO-FAS: SISO-FAS activates the port with the strongest channel before expressing its rate and outage probability.The received signal is described at each fluid-antenna port, and optimal port selection is based on channel strength.
  • SISO-FAS: At high SNR, the SISO-FAS outage probability is characterized as a function of the SNR threshold.The supplied result passage identifies the high-SNR outage expression without reproducing its full equation.
  • Performance Comparison: Fig. 5 compares SISO-FAS, MIMO-FAS, and dual-fluid-antenna configurations against SNR, with fluid antennas placed at one or both ends.The dual-FAS curves use fluid antennas at both transmitter and receiver, while FAS curves use one fluid antenna.
  • MIMO-FAS: MIMO-FAS activates multiple ports and represents the input covariance through beamforming and power allocation matrices.The rate formulation incorporates the effective channel, transmit signal, AWGN, and a transmit-SNR constraint.
  • MIMO-FAS: MIMO-FAS rate maximization jointly optimizes port selection, transmit and receive beamforming, and power allocation, but exhaustive search, SVD, and waterfilling incur high complexity.These methods are respectively used for port selection, beamforming, and power allocation.

2) Theoretical Performance:

The surveyed results show that FAS can improve channel capacity and wideband throughput, while port selection can also add spectral efficiency in index-modulation systems.

  • Theoretical Performance: FAS-assisted channel capacity improves over traditional FPA systems, with further enhancement when fluid antennas are deployed at both transmitter and receiver.The comparison includes SISO-FAS, MIMO-FAS, and dual-FAS configurations under the stated antenna specifications.
  • Theoretical Performance: FAS improves throughput in 20 MHz wideband simulations across different modulation and coding schemes compared with traditional FPA systems.The receiver-side FAS configuration uses 4 × 4 ports distributed over a 0.5λ × 0.5λ plane.
  • RIS-aided FAS: RIS-aided FAS adjusts RIS reflecting phases to maximize the combined channel gain before the user selects the strongest FAS port.The phase choice aligns the RIS-assisted paths, and port selection follows the resulting combined gain.
  • FAS-aided Index Modulation: FAS-aided index modulation selects fluid-antenna ports to encode information, and its spectral efficiency includes an additional contribution from index modulation.The receiver detects both the transmitted symbol and the selected port index under the stated channel model.

D. Multi-user FAMA and its Variants

FAMA uses fluid-antenna port selection to manage multi-user interference, with performance depending on update speed, user count, SINR threshold, and spectral efficiency.

  • Transmission Model: FAMA selects the port or ports with the highest SINR at each user, with the SINR definition differing between slow and fast variants.Slow FAMA updates less frequently, whereas fast FAMA evaluates instantaneous SINR.
  • Theoretical Performance: Theoretical FAMA analysis defines outage probability and multiplexing gain, with upper bounds derived for both slow and fast schemes.The multiplexing gain measures capacity scaling under a fixed transmitted rate per user.
  • Theoretical Performance: The simplified FAMA channel model can produce artificially optimistic performance predictions, while the more accurate model is substantially more complicated to analyze.The paper notes that outage results for more accurate models have been derived for selected FAMA cases.
  • Theoretical Performance: Slow-FAMA multiplexing gain decreases as the SINR threshold or user count increases and approaches 0 for U = 10 or 30 because of interference.Fast FAMA retains considerable multiplexing gain even with many users, whereas slow FAMA is stronger when the user count is relatively low.
  • Link-level Simulation Results: Link-level simulations show that slow FAMA provides great multiplexing gain at low spectral efficiency, with gain decreasing as spectral efficiency increases.The simulated metric is recalculated as GFAMA = (1 − BLER)^U.

4) Variants of FAMA:

This section describes port-selection and symbol-detection procedures for FAMA variants, including quadrant-focused processing and a discrete-time SISO-FAS communication model. It also motivates networking techniques for QoS and resource efficiency.

  • Port selection: Ports are selected from positive and negative in-phase channel sets using a minimum in-phase threshold and a stated selection criterion.The final port set includes only ports satisfying the selection condition.
  • Symbol detection: The selected ports’ in-phase and quadrature contributions are aggregated before the received symbol is detected.The detection process follows the aggregation of the in-phase and quadrant components.
  • Symbol detection: Repeating the process with emphasis on the quadrature component further improves detection quality, and CUMA can support hundreds of users per channel use.The stated user-support capability depends on sophisticated port selection.
  • System model: The SISO-FAS model links a point-to-point communication system with a discrete-time queueing model, while practical implementations may use pixels or metamaterials instead of fluidic antennas.The model is presented as an upper-layer communication system representation.
  • Networking perspective: Networking techniques are presented as necessary for efficient, reliable, and secure FAS operation, with QoS, power allocation, and content placement identified as focal areas.The section emphasizes that FAS networking research remains in preliminary stages.

1) QoS Modelling in Mobile Networks:

This section models FAS QoS through a discrete-time queue with random arrivals and service over a block-fading channel. Effective bandwidth and effective capacity are used to analyze delay distributions, including delay-outage probability, expectation, and variance.

  • Queueing model: The model defines bandwidth B, transmit power Pt, noise spectral density N0, and slot duration Ts to characterize the discrete-time system.Path loss and noise variance are also included in the system description.
  • Queueing model: Upper-layer data enter a FIFO buffer and are transmitted over a block-fading channel whose gains remain constant during each time slot.The arrival process and transmission capability are represented by slot-indexed random variables.
  • Traffic and service: Arrivals A[n] are modeled as i.i.d. exponential random variables, while service S[n] follows a fixed distribution and Q[n] denotes backlog size.The average data rate is defined from the exponential arrival parameter and slot duration.
  • Capacity model: With instantaneous FAS channel gain known at the transmitter, slot capacity C is related to service S and provides the basis for effective-capacity analysis.The channel power-gain distribution and effective capacity are specified for an N-port FAS.
  • Delay distribution: Under Gartner–Ellis assumptions and a unique positive QoS exponent, delay CCDF, expectation, and variance can be approximated analytically.The delay-distribution analysis uses effective bandwidth and effective capacity models.
  • Numerical evaluation: The simulation evaluates delay-outage probability versus maximum delay for N = 1, 5, and 50 using 10^6 backlog samples under a specified 1D-FAS setting.The x-axis is delay bound in milliseconds and the y-axis is violation probability on a logarithmic scale.

B. Power Allocation

This section surveys FAS power allocation objectives, solution methods, and application settings, then examines energy-efficiency optimization under delay-outage constraints. The case study reports how delay bounds, port count, and FAS size affect optimal EE.

  • Objectives and constraints: FAS power allocation mainly targets energy-efficiency or sum-rate maximization under application-specific constraints.Common constraints include statistical delay bounds, total power budgets, and limits on activated FAS ports.
  • Solution methodologies: The surveyed solution methods comprise convex optimization and decomposition, learning-based approaches, and hybrid techniques.Decomposition can turn non-convex joint problems into tractable subproblems, while deep reinforcement learning addresses high-dimensional state spaces.
  • Emerging applications: Emerging applications jointly optimize variables such as UAV altitude, FAS port selection, and NOMA power coefficients, while FAS mitigates small-scale fading in UAV-relay scenarios.The section also considers power allocation in energy-harvesting settings.
  • QoS-constrained optimization: Under a target delay-outage constraint {Dmax, ϵ}, transmit power is selected while other parameters remain fixed to optimize energy efficiency.The target QoS exponent is obtained from the delay constraint, and the optimal power satisfies an equation solvable by numerical root finding.
  • Case study: EE increases as Dmax is extended and the DOP constraint is relaxed, with a 50-port FAS achieving about 2.02 to 16.11 times the EE of an FPA system at Dmax = 1 or 10 ms.As Dmax approaches infinity, the optimal transmit power converges to the stable-system condition R = E[C].
  • Case study: Under {Dmax = 5 ms, ϵ = 0.02}, EE increases with port count but gains become less pronounced at large N, while increasing FAS size yields marginal improvements.The EE definition includes constant circuit power and transmit power in total transmission power.

C. Network Technologies in FAS HetNets

This section presents FAS as a spatially agile technology for dense HetNets and content-centric networks, where dynamic port selection can support interference management and content delivery. It models probabilistic caching, user association, channel behavior, and port selection in FAS-assisted HetNets.

  • FAS in HetNets: FAS dynamically repositions antenna elements over a predefined surface, providing spatial diversity, signal robustness, and control over propagation conditions.In HetNets, the surveyed applications include dynamic beam steering, adaptive user association, and spatial interference mitigation.
  • FAS in HetNets: FAS-enabled HetNets face added complexity because user association is less predictable, CSI acquisition and tracking become more frequent, and spatial agility affects scheduling and resource allocation.Content-centric networking is highlighted as an application where proactive caching can reduce latency and backhaul load.
  • Content-centric networking: The content-delivery model uses SBS tiers with limited storage, popularity-based caching, nearest-SBS association, and a finite content set connected through a macro-cell backhaul.The model focuses on content delivery and placement in a multi-user CCN-enabled HetNet.
  • Content placement: Each SBS independently caches content l with probability ql, subject to a total caching constraint, with more popular contents assigned higher caching likelihood.The probabilistic caching vector satisfies a storage limit of K contents across L available contents.
  • Network model: A typical FAS-equipped UE has N = N1 × N2 ports and size W = W1λ × W2λ, while SBS locations follow an independent homogeneous Poisson point process.Each UE communicates with one SBS per time slot, and SBSs caching content l have intensity qlµS.
  • Channel and port selection: The HetNet channel model includes path loss and combined noise or weak interference, with the best port selected by maximizing received SNR.The port-channel dependence is modeled through covariance and copula-based representations under rich scattering.

2) Performance Analysis:

The performance analysis models SCDP and CDD for FAS-enabled content delivery and shows that spatial adaptability improves reliability and reduces delay relative to fixed-position antennas. Gains depend on SNR threshold, caching probability, antenna configuration, SBS density, and retransmission rounds.

  • SCDP analysis: FAS improves SCDP over FPA by dynamically selecting antenna positions that optimize instantaneous channel gain and reduce blockage effects.The benefit is especially relevant in dense deployments.
  • SCDP analysis: Increasing both W and N improves SCDP, but increasing N beyond a certain point can saturate gains because of stronger spatial correlation and greater hardware complexity.This creates a practical trade-off between spatial diversity, capacity, and implementation complexity.
  • SCDP analysis: SCDP increases monotonically with caching probability q_l for both FAS and FPA, while FAS provides a larger gain when cached content is available nearby.Increasing SBS density μ_S also improves SCDP by reducing UE–SBS distance and enhancing SNR.
  • CDD analysis: At γ = 15 dB, FAS reduces CDD to approximately 1.72 ms versus about 2.6 ms for FPA, a reduction of 33.85%.The comparison uses an FAS UE with N = 3 × 3 and W = 2λ.
  • CDD analysis: FAS-equipped UEs achieve significantly lower CDD than FPA-based UEs across varying SNR thresholds.The analysis attributes the reduction to improved transmission reliability from FAS deployment.
  • CDD analysis: Increasing ARQ rounds raises CDD because additional retransmission attempts increase delivery time.Lower caching probability also increases delay because requested content is less likely to be available at the nearest SBS.

V. Challenges and Open Issues

The survey identifies practical barriers to deploying FAS, including incomplete channel models, high CSI and hardware demands, uncertain energy efficiency, security and privacy risks, and integration costs. It concludes that further networking research and realistic measurements are needed to mature the technology.

  • Challenges and open issues: Accurate FAS electromagnetic models must balance computational complexity with real-world accuracy across far-field and near-field conditions.The need becomes more pressing as operating frequencies increase and spatial resolution becomes finer.
  • Challenges and open issues: Continuous channel estimation for dynamic reconfiguration creates CSI overhead, while large-scale configurations still require many observation ports for accurate recovery.Reducing CSI overhead without compromising performance remains an open concern.
  • Challenges and open issues: Empirical FAS channel models remain scarce because they require extensive measurements, while current prototypes face durability and miniaturization problems.Existing channel-model research is dominated by theoretical frameworks.
  • Challenges and open issues: Realistic energy-efficiency analysis must account for dynamic reconfiguration power and multi-cell operation rather than treating circuit power as constant.Existing evaluation has focused on single-cell environments.
  • Challenges and open issues: FAS reconfigurability introduces potential beam-hijacking and location-privacy vulnerabilities that require encryption, beam nulling, and anonymization research.The survey characterizes these protections as still under development or requiring further investigation.
  • Challenges and open issues: High infrastructure costs, new CSI feedback requirements, HARQ integration challenges, and scarce benchmark datasets constrain practical FAS development.The scarcity of datasets is linked to the early stage of FAS hardware development and limited practical channel measurements.
  • Conclusion: The survey covers FAS application scenarios, physical-layer fundamentals, QoS, power allocation, and content placement while identifying networking research directions.Content placement in FAS-based HetNets is discussed as a way to improve efficiency by reducing delivery delays.
Loading 2506.13317v2…