arxiv: 2604.16639 · v1 · submitted 2026-04-17 · 💻 cs.IT · math.IT

Recognition: unknown

Beyond Covariance: Generative Spatial Correlation Modeling and Channel Interpolation for Fluid Antenna Systems

Zhentian Zhang , Hao Jiang , Kai-Kit Wong , Hyundong Shin , Ross Murch

Authors on Pith no claims yet

Pith reviewed 2026-05-10 06:57 UTC · model grok-4.3

classification 💻 cs.IT math.IT

keywords fluid antenna systemschannel interpolationautoregressive modelGauss-Markov processMMSE estimationKalman filteringspatial correlationchannel state information

0 comments

The pith

Fluid antenna systems model spatially sampled channels as autoregressive Gauss-Markov processes to derive optimal MMSE interpolation from sparse observations.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper shows that representing the channel across fluid antenna ports as a pth-order autoregressive Gauss-Markov process captures spatial correlation in a generative way. This model lets the authors derive the globally optimal minimum mean-square error estimator and a tight lower bound on how many observations are needed to keep reconstruction error below a target level. They then use the state-space structure of the AR process to build a Kalman filter and smoother that achieves this optimal performance at strictly linear complexity in the number of ports. A reader would care because conventional covariance methods lack generalization and scale poorly when the port count grows large, while this approach supplies both theoretical limits and a practical algorithm.

Core claim

By adopting the pth-order AR Gauss-Markov process as the generative model for the spatially sampled channel, the paper derives the globally optimal MMSE estimator for interpolating CSI across all ports from a small number of observations, proves a tight lower bound on the number of observations required to meet any prescribed error, and realizes the estimator with a Kalman filter/smoother whose complexity is linear in the number of ports.

What carries the argument

The pth-order autoregressive Gauss-Markov process, which generates the channel values recursively from past samples and supplies an exact state-space representation for Kalman-based MMSE estimation.

If this is right

The MMSE estimator is globally optimal under the AR(p) assumption for any choice of observation locations.
A lower bound exists on the minimum number of observations needed to keep reconstruction error below any given threshold.
Kalman filtering and smoothing achieve the optimal MMSE performance with O(N) complexity.
Increasing the AR order p trades higher model accuracy for greater computational cost in a controlled way.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

The same AR modeling approach could be applied to other dense antenna arrays where port switching occurs, not only fluid antennas.
The linear complexity suggests the method could run in real time even when the number of ports reaches thousands.
One could test the tightness of the observation lower bound by generating synthetic channels exactly matching the AR model and counting how many samples are actually required.

Load-bearing premise

The spatially sampled channel can be represented as a pth-order autoregressive Gauss-Markov process.

What would settle it

Channel measurements from a fluid antenna prototype that cannot be fitted by any AR(p) model to within the reconstruction error predicted by the derived lower bound.

Figures

Figures reproduced from arXiv: 2604.16639 by Hao Jiang, Hyundong Shin, Kai-Kit Wong, Ross Murch, Zhentian Zhang.

**Figure 1.** Figure 1: Illustration of the proposed AR(p) Gauss-Markov correlation modeling under different order coefficients p ∈ {5, 10, 20} with N = 200, W = 5, burn-in length B = 5N, indicating there is an optimal AR order under fixed (W, N). All CDFs are generated via 3 × 104 Monte-Carlo samples. C. Optimal AR Order p ⋆ for Fixed (W, N) Define the AR polynomial in the z-domain A(z) = 1 − Xp i=1 αiz −i . (19) A standard stab… view at source ↗

**Figure 2.** Figure 2: Illustration of the maximum error of the proposed AR( [PITH_FULL_IMAGE:figures/full_fig_p006_2.png] view at source ↗

**Figure 3.** Figure 3: Illustration of the analytically structured CDF by ( [PITH_FULL_IMAGE:figures/full_fig_p008_3.png] view at source ↗

**Figure 4.** Figure 4: Correlation modeling comparison between the AR( [PITH_FULL_IMAGE:figures/full_fig_p011_4.png] view at source ↗

**Figure 5.** Figure 5: One realization of channel interpolation with diffe [PITH_FULL_IMAGE:figures/full_fig_p012_5.png] view at source ↗

**Figure 7.** Figure 7: Illustration of empirical and theoretical results fo [PITH_FULL_IMAGE:figures/full_fig_p012_7.png] view at source ↗

read the original abstract

Fluid antenna systems (FAS) enable unprecedented spatial diversity within a compact form factor by flexibly switching among high-density antenna ports. To activate this capability, channel state information (CSI) over the ports is required, which implies high estimation overhead because the number of ports is usually very large. Conventional estimation schemes tend to first estimate the CSI for a small number of ports and then infer the CSI for the remaining antenna ports by interpolation exploiting correlation characteristics. However, existing correlation-based techniques lack generalization ability, and the fundamental limits of interpolating the CSI from sparse observations remain poorly understood. This paper adopts a generative modeling framework for characterizing the channel correlation among the FAS ports that departs fundamentally from covariance-descriptive models. Specifically, we represent the spatially sampled channel as a $p$th-order autoregressive (AR) Gauss-Markov process, which provides a principled and tunable tradeoff between model complexity and approximation accuracy via the AR order. In so doing, we can characterize the limits of channel interpolation by deriving the globally optimal minimum mean-square error (MMSE) estimator and establishing a tight lower bound on the minimum number of observations required to meet a prescribed reconstruction error. To reduce the complexity of MMSE estimation, we then exploit the state-space structure due to the ${\rm AR}(p)$ model and develop a Kalman filtering/smoothing-based interpolation algorithm. The resulting method attains the optimal MMSE performance with strictly linear complexity $\mathcal{O}(N)$ with $N$ denoting the number of ports, resulting in a scalable, efficient, and theoretically grounded framework for practical FAS channel reconstruction.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Desk Editor's Note private letter to a colleague

The paper models FAS channels as an AR(p) Gauss-Markov process to get exact MMSE interpolation limits plus a Kalman smoother that hits optimality at linear cost.

read the letter

The punchline is that switching to a generative AR(p) model for the spatially sampled channel gives both a clean lower bound on the number of observations needed for a target error and an O(N) algorithm that achieves the MMSE estimator. That combination is the real step forward for fluid antennas, where port counts are high and conventional covariance interpolation does not scale well or come with provable limits. The derivation uses the state-space structure of the AR process, so the Kalman recursions are optimal by construction and the complexity stays linear for fixed p. This is useful because it turns the interpolation problem into something that can be run in real time on compact devices. The bound on observations is also direct from the model covariance, which is a nice theoretical result even if it stays inside the assumed process. The main limitation is that the whole story rests on how well real FAS channels match an AR(p) Gauss-Markov process. The paper treats p as a free parameter that trades accuracy for complexity, but without strong empirical checks against measured or ray-traced data the practical value stays conditional. If the full manuscript only has synthetic validation, that part needs tightening before it influences 6G designs. Readers working on movable-antenna or fluid-antenna CSI acquisition will get the most out of it; the framework is concrete enough that people building simulators or testbeds could adapt the Kalman step quickly. It is worth sending to peer review because the internal logic is consistent and the complexity claim is verifiable, even if reviewers will press on the modeling assumption and the choice of p.

Referee Report

2 major / 2 minor

Summary. The paper proposes modeling the spatially sampled FAS channel as a pth-order autoregressive (AR) Gauss-Markov process to enable generative characterization of spatial correlation. Under this model it derives the globally optimal MMSE estimator for interpolating CSI from sparse port observations, establishes a tight lower bound on the minimum number of observations needed to achieve a target reconstruction error, and shows that Kalman smoothing realizes the MMSE solution at strictly linear O(N) complexity.

Significance. If the derivations are correct, the work supplies a theoretically grounded alternative to covariance-only interpolation methods, with explicit optimality guarantees and a tunable complexity-accuracy tradeoff via the AR order. The use of standard linear-Gaussian state-space techniques (Kalman recursions) to attain both optimality and linear scaling is a clear strength, as is the derivation of a model-based lower bound on observation count.

major comments (2)

[AR model definition and bound derivation] The AR order p is introduced as a tunable parameter providing a complexity-accuracy tradeoff, yet the claimed tight lower bound on the number of observations and the MMSE optimality are derived conditionally on a fixed p. The manuscript should clarify (e.g., in the section deriving the bound) whether the bound expression depends on p and how p is to be chosen in practice without introducing post-hoc fitting that affects the reported performance guarantees.
[Kalman filtering/smoothing algorithm] The abstract asserts that the Kalman smoother attains the globally optimal MMSE estimator, but without the explicit state-space equations or the proof that the smoother matches the MMSE solution for the AR(p) process, it is difficult to verify that no additional assumptions (e.g., on initial conditions or boundary handling) are required for exact optimality.

minor comments (2)

[System model] Notation for the observation model and noise statistics should be introduced consistently before the MMSE derivation to avoid ambiguity in the subsequent bounds.
[Numerical results] Any numerical results or figures comparing the proposed method to existing covariance-based interpolators should be accompanied by explicit statements of the AR order p used and the channel generation process.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the thorough review and positive recommendation for minor revision. We address each major comment point-by-point below and will incorporate clarifications into the revised manuscript.

read point-by-point responses

Referee: [AR model definition and bound derivation] The AR order p is introduced as a tunable parameter providing a complexity-accuracy tradeoff, yet the claimed tight lower bound on the number of observations and the MMSE optimality are derived conditionally on a fixed p. The manuscript should clarify (e.g., in the section deriving the bound) whether the bound expression depends on p and how p is to be chosen in practice without introducing post-hoc fitting that affects the reported performance guarantees.

Authors: The lower bound on the number of observations and the MMSE optimality are derived for a given fixed AR order p, as the generative model is the AR(p) process. The bound expression explicitly depends on p because the correlation structure (and thus the information provided by each observation) varies with the model order. In the revised manuscript, we will add a statement in the bound derivation section clarifying this dependence. For the practical choice of p, we emphasize that all theoretical guarantees in the paper are conditional on the channel obeying the AR(p) model. In practice, p can be determined from the physical propagation characteristics or selected via information criteria on a calibration dataset independent of the test channels. This does not constitute post-hoc fitting that invalidates the guarantees, as the optimality and bound hold exactly under the assumed model; we will include a brief discussion of this point to address the concern. revision: partial
Referee: [Kalman filtering/smoothing algorithm] The abstract asserts that the Kalman smoother attains the globally optimal MMSE estimator, but without the explicit state-space equations or the proof that the smoother matches the MMSE solution for the AR(p) process, it is difficult to verify that no additional assumptions (e.g., on initial conditions or boundary handling) are required for exact optimality.

Authors: We agree that making the connection explicit will improve verifiability. The AR(p) Gauss-Markov process has a canonical state-space realization with state vector of dimension p containing the current and lagged channel values, driven by a companion transition matrix and scalar process noise. Under this linear Gaussian state-space model, the Kalman smoother computes the exact posterior means, which coincide with the MMSE interpolator derived earlier. Initial conditions are set using the stationary covariance of the AR process, and the smoother handles boundaries without approximation by running the forward-backward recursions over the entire sequence. In the revised manuscript, we will insert the state-space equations and a short equivalence proof in the Kalman section to confirm exact optimality under the model assumptions. revision: yes

Circularity Check

0 steps flagged

No significant circularity identified

full rationale

The paper explicitly adopts an AR(p) Gauss-Markov generative model as a modeling choice that trades off complexity and accuracy via the tunable order p. From this assumption it derives the MMSE estimator (via the standard state-space Kalman smoother, which is known to be optimal for linear-Gaussian AR processes) and a lower bound on the number of observations needed for a target error (which follows directly from the model's covariance). The O(N) complexity is likewise a standard property of the Kalman recursions for fixed p. None of these steps reduces by construction to a fitted parameter, a self-citation chain, or a renaming of an input; all results are conditional on the stated model assumption and are therefore self-contained.

Axiom & Free-Parameter Ledger

1 free parameters · 1 axioms · 0 invented entities

The central claim rests on modeling the channel as an AR(p) Gauss-Markov process; p is the only explicit free parameter, and the Gauss-Markov property is the key domain assumption.

free parameters (1)

AR order p
Tunable integer controlling the memory depth and thus the tradeoff between model fidelity and complexity in the generative channel representation.

axioms (1)

domain assumption Spatially sampled FAS channel is a pth-order autoregressive Gauss-Markov process
Invoked to enable the generative framework, MMSE derivation, and state-space Kalman structure.

pith-pipeline@v0.9.0 · 5595 in / 1316 out tokens · 38263 ms · 2026-05-10T06:57:18.603220+00:00 · methodology

discussion (0)

Reference graph

Works this paper leans on

52 extracted references · 8 canonical work pages · 2 internal anchors

[1]

Layered space-time architecture for wi reless communication in a fading environment when using multi-element antennas,

G. J. Foschini, “Layered space-time architecture for wi reless communication in a fading environment when using multi-element antennas,” Bell Labs Tech. J. , vol. 1, no. 2, pp. 41–59, Autumn 1996

1996
[2]

Capacity of multi-antenna Gaussian channe ls,

E. Telatar, “Capacity of multi-antenna Gaussian channe ls,” Euro. Trans. Telecom- mun., vol. 10, no. 6, pp. 585–595, 1999

1999
[3]

O ptimizing the spectral efﬁciency of multiuser MIMO smart antenna systems ,

K. K. Wong, R. D. Murch, R. S.-K. Cheng and K. B. Letaief, “O ptimizing the spectral efﬁciency of multiuser MIMO smart antenna systems ,” in Proc. IEEE Wirel. Commun. Netw. Conf. (WCNC) , vol. 1, pp. 426–430, 23-28 Sept. 2000, Chicago, IL, USA

2000
[4]

Performance enh ancement of multiuser MIMO wireless communication systems,

K. K. Wong, R. D. Murch and K. B. Letaief, “Performance enh ancement of multiuser MIMO wireless communication systems,” IEEE Trans. Commun. , vol. 50, no. 12, pp. 1960–1970, Dec. 2002

1960
[5]

Massive MIMO for next generation wireless systems,

E. G. Larsson, O. Edfors, F. Tufvesson and T. L. Marzetta, “Massive MIMO for next generation wireless systems,” IEEE Commun. Mag. , vol. 52, no. 2, pp. 186–195, Feb. 2014

2014
[6]

Spectral efﬁciency of precoded 5G-NR in si ngle and multi- user scenarios under imperfect channel knowledge: A compre hensive guide for implementation,

D. A. U. Villalonga, H. OdetAlla, M. J. Fern´ andez-Getin o Garc´ ıa, and A. Flizikowski, “Spectral efﬁciency of precoded 5G-NR in si ngle and multi- user scenarios under imperfect channel knowledge: A compre hensive guide for implementation,” Electronics, vol. 11, no. 24, p. 4237, Dec. 2022

2022
[7]

A tutorial on extremely large-scale MIMO for 6G: Fundamen tals, signal processing, and applications,

Z. Wang et al. , “A tutorial on extremely large-scale MIMO for 6G: Fundamen tals, signal processing, and applications,” IEEE Commun. Surv. & Tut. , vol. 26, no. 3, pp. 1560–1605, thirdquarter 2024

2024
[8]

PAPR analysis and mitigati on algorithms for beamforming MIMO OFDM systems,

Y .-C. Hung and S.-H. L. Tsai, “PAPR analysis and mitigati on algorithms for beamforming MIMO OFDM systems,” IEEE Trans. Wireless Commun. , vol. 13, no. 5, pp. 2588–2600, May 2014

2014
[9]

J. T. Bernhard, Reconﬁgurable antennas , Synthesis Lectures on Antennas, 2007

2007
[10]

Per formance limits of ﬂuid antenna systems,

K. K. Wong, A. Shojaeifard, K.-F. Tong and Y . Zhang, “Per formance limits of ﬂuid antenna systems,” IEEE Commun. Lett. , vol. 24, no. 11, pp. 2469–2472, Nov. 2020

2020
[11]

Flu id antenna systems,

K. K. Wong, A. Shojaeifard, K.-F. Tong and Y . Zhang, “Flu id antenna systems,” IEEE Trans. Wireless Commun. , vol. 20, no. 3, pp. 1950–1962, Mar. 2021

1950
[12]

A tutorial on ﬂuid antenna system for 6G networks: Encom- passing communication theory, optimization methods and ha rdware designs,

W. K. New et al. , “A tutorial on ﬂuid antenna system for 6G networks: Encom- passing communication theory, optimization methods and ha rdware designs,” IEEE Commun. Surv. Tuts. , vol. 27, no. 4, pp. 2325–2377, Aug. 2025

2025
[13]

A contemporary survey on ﬂuid antenna systems: Fundament als and networking perspectives,

H. Hong et al. , “A contemporary survey on ﬂuid antenna systems: Fundament als and networking perspectives,” IEEE Trans. Netw. Sci. Eng. , vol. 13, pp. 2305–2328, 2026

2026
[14]

Fluid antennas: Reshaping intrinsic properties for ﬂexi ble radiation characteristics in intelligent wireless networks,

W.-J. Lu et al., “Fluid antennas: Reshaping intrinsic properties for ﬂexi ble radiation characteristics in intelligent wireless networks,” IEEE Commun. Mag. , vol. 63, no. 5, pp. 40–45, May 2025

2025
[15]

Fluid antenna systems: Redeﬁning reconﬁgurable wireles s communications,

W. K. New et al. , “Fluid antenna systems: Redeﬁning reconﬁgurable wireles s communications,” IEEE J. Sel. Areas Commun. , vol. 44, pp. 1013–1044, 2026

2026
[16]

Fluid antenna systems enabling 6G: Principles, applica- tions, and research directions,

T. Wu et al. , “Fluid antenna systems enabling 6G: Principles, appli- cations, and research directions,” IEEE Wireless Commun. , Early Access, doi:10.1109/MWC.2025.3629597, 2025

work page doi:10.1109/mwc.2025.3629597 2025
[17]

Historical review of fluid antenna and movable antenna,

L. Zhu and K. K. Wong, “Historical review of ﬂuid antenna s and movable antennas,” arXiv preprint , arXiv:2401.02362v2, 2024

work page arXiv 2024
[18]

Design and experimental validation of mmWave surface wav e enabled ﬂuid antennas for future wireless communications,

Y . Shen et al. , “Design and experimental validation of mmWave surface wav e enabled ﬂuid antennas for future wireless communications, ” to appear in IEEE Antennas & Wireless Propag. Lett. , doi:10.1109/LAWP .2026.3657059, 2026

work page doi:10.1109/lawp 2026
[19]

Electromagnetically reconﬁgurable ﬂuid antenna system for wire- less communications: Design, modeling, algorithm, fabric ation, and experiment,

R. Wang et al. , “Electromagnetically reconﬁgurable ﬂuid antenna system for wire- less communications: Design, modeling, algorithm, fabric ation, and experiment,” IEEE J. Select. Areas Commun. , vol. 44, pp. 1464–1479, 2026

2026
[20]

A novel pixel-based reconﬁgurable antenna applied in ﬂui d antenna systems with high switching speed,

J. Zhang et al., “A novel pixel-based reconﬁgurable antenna applied in ﬂui d antenna systems with high switching speed,” IEEE Open J. Antennas & Propag. , vol. 6, no. 1, pp. 212–228, Feb. 2025

2025
[21]

Wideban d pixel-based ﬂuid antenna system: An antenna design for smart city,

B. Liu, T. Wu, K. K. Wong, H. Wong, and K. F. Tong, “Wideban d pixel-based ﬂuid antenna system: An antenna design for smart city,” IEEE Internet Things J. , vol. 13, no. 4, pp. 6850–6862, Feb. 2026

2026
[22]

Fluid antenna systems enabled by reconﬁgurable holograp hic surfaces: Beamforming design and experimental validation ,

S. Zhang et al. , “Fluid antenna systems enabled by reconﬁgurable holograp hic surfaces: Beamforming design and experimental validation ,” IEEE J. Select. Areas Commun., vol. 44, pp. 1417–1431, 2026

2026
[23]

P rogrammable meta- ﬂuid antenna for spatial multiplexing in fast ﬂuctuating ra dio channels,

B. Liu, K.-F. Tong, K. K. Wong, C.-B. Chae, and H. Wong, “P rogrammable meta- ﬂuid antenna for spatial multiplexing in fast ﬂuctuating ra dio channels,” Optics Express, vol. 33, no. 13, pp. 28898–28915, 2025

2025
[24]

An information- theoretic characterization of MIMO-FAS: Optimization, di versity-multiplexing tradeoff and q-outage capacity,

W. K. New, K. K. Wong, H. Xu, K.-F. Tong and C.-B. Chae, “An information- theoretic characterization of MIMO-FAS: Optimization, di versity-multiplexing tradeoff and q-outage capacity,” IEEE Trans. Wireless Commun. , vol. 23, no. 6, pp. 5541–5556, Jun. 2024

2024
[25]

A Gaussian copula approach to the performance analysis of ﬂuid antenna systems,

F. Rostami Ghadi et al. , “A Gaussian copula approach to the performance analysis of ﬂuid antenna systems,” IEEE Trans. Wireless Commun. , vol. 23, no. 11, pp. 17573–17585, Nov. 2024

2024
[26]

A new spatial block- correlation model for ﬂuid antenna systems,

P . Ram´ ırez-Espinosa, D. Morales-Jimenez and K. K. Won g, “A new spatial block- correlation model for ﬂuid antenna systems,” IEEE Trans. Wireless Commun. , vol. 23, no. 11, pp. 15829–15843, Nov. 2024

2024
[27]

Finite-blocklength ﬂuid antenna systems with spatial bl ock- correlation channel model,

Z. Zhang et al. , “Finite-blocklength ﬂuid antenna systems with spatial bl ock- correlation channel model,” IEEE Wireless Commun. Lett. , vol. 15, pp. 1911-1915, 2026

1911
[28]

Fluid antenna multiple access ,

K. K. Wong and K. F. Tong, “Fluid antenna multiple access ,” IEEE Trans. Wireless Commun., vol. 21, no. 7, pp. 4801–4815, Jul. 2022

2022
[29]

On fundamental limits of slow-ﬂuid antenna multiple acce ss for unsourced random access,

Z. Zhang, et al. , “On fundamental limits of slow-ﬂuid antenna multiple acce ss for unsourced random access,” IEEE Wireless Commun. Lett. , vol. 14, no. 11, pp. 3455–3459, Nov. 2025

2025
[30]

Cramer-Rao Bounds for Activity Detection in Conventional and Fluid Antenna Systems

Z. Zhang, K. K. Wong, H. Jiang, C. Masouros, and C.-B. Cha e, “On fundamental limits of transmission activity detection in ﬂuid antenna s ystems,” arXiv preprint , arXiv:2602.11901, 2026

work page internal anchor Pith review Pith/arXiv arXiv 2026
[32]

O n fundamental limits for ﬂuid antenna-assisted integrated sensing and communic ations for unsourced random access,

Z. Zhang, K. K. Wong, J. Dang, Z. Zhang, and C.-B. Chae, “O n fundamental limits for ﬂuid antenna-assisted integrated sensing and communic ations for unsourced random access,” IEEE J. Sel. Areas Commun. , vol. 44, pp. 136–149, 2026

2026
[33]

Finite-Aperture Fluid Antenna Array Design: Analysis and Algorithm

Z. Zhang et al.,“Finite-aperture ﬂuid antenna array design: Analysis and algorithm,” arXiv preprint , arXiv:2601.18471, 2026

work page internal anchor Pith review Pith/arXiv arXiv 2026
[34]

Channel estimation for FAS-assisted multiuser mmWave sy stems,

H. Xu et al. , “Channel estimation for FAS-assisted multiuser mmWave sy stems,” IEEE Commun. Lett. , vol. 28, no. 3, pp. 632–636, Mar. 2024

2024
[35]

Channel estimation and reconstruction in ﬂuid antenna sy stem: Oversampling is essential,

W. K. New et al. , “Channel estimation and reconstruction in ﬂuid antenna sy stem: Oversampling is essential,” IEEE Trans. Wireless Commun., vol. 24, no. 1, pp. 309– 322, Jan. 2025

2025
[36]

Low-complexity CSI acquisition exploiting geographica l diversity in ﬂuid antenna system,

Z. Zhang et al., “Low-complexity CSI acquisition exploiting geographica l diversity in ﬂuid antenna system,” arXiv preprint , arXiv:2509.08598, 2025

work page arXiv 2025
[37]

Neur al networks-enabled channel reconstruction for ﬂuid antenna systems: A data-dr iven approach,

H. Liang, Z. Zhang, J. Dang, H. Jiang, and Z. Zhang, “Neur al networks-enabled channel reconstruction for ﬂuid antenna systems: A data-dr iven approach,” accepted by WCNC 2026, available:arXiv:2511.14520, 2025

work page arXiv 2026
[38]

Joint activity detection and channel estimation for ﬂuid antenna systems exploiting geographical and angular infor mation,

Z. Zhang et al. , “Joint activity detection and channel estimation for ﬂuid antenna system exploiting geographical and angular information,” IEEE J. Sel. Topics Signal Process., doi:10.1109/JSTSP .2026.3673148, 2025

work page doi:10.1109/jstsp 2026
[39]

W. C. Jakes and D. C. Cox, Microwave Mobile Communications . Hoboken, NJ, USA: Wiley, 1994

1994
[40]

A statistical theory of mobile-radio rec eption,

R. H. Clarke, “A statistical theory of mobile-radio rec eption,” Bell Syst. Tech. J. , vol. 47, no. 6, pp. 957–1000, Jul.-Aug. 1968

1968
[41]

When is Clarke’s approxima tion valid?,

T. W. C. Brown and U. M. Ekpe, “When is Clarke’s approxima tion valid?,” IEEE Antennas & Propag. Mag. , vol. 52, no. 3, pp. 171–181, Jun. 2010

2010
[42]

P . J. Brockwell and R. A. Davis, Time series: Theory and methods , 2nd ed., New Y ork: Springer, 1991

1991
[43]

G. E. P . Box, G. M. Jenkins, and G. C. Reinsel, Time series analysis: Forecasting and control, 3rd ed., Englewood Cliffs, NJ: Prentice-Hall, 1994

1994
[45]

Y ule-Walker equations us ing higher order statistics for nonlinear autoregressive model,

U. Libal and K. H. Johansson, “Y ule-Walker equations us ing higher order statistics for nonlinear autoregressive model,” in Proc. Sig. Proc. Symp. (SPSympo) , pp. 227– 231, 17-19 Sept. 2019, Krakow, Poland

2019
[46]

P . J. Brockwell and R. A. Davis, Introduction to time series and forecasting , 3rd ed., Springer, 2016

2016
[47]

Durbin and S

J. Durbin and S. J. Koopman, Time series analysis by state space methods , 2nd ed. Oxford University Press, 2012

2012
[48]

Generalized Markov mod els for real-time modeling of continuous systems,

D. P . Filev and I. Kolmanovsky, “Generalized Markov mod els for real-time modeling of continuous systems,” IEEE Trans. Fuzzy Syst. , vol. 22, no. 4, pp. 983– 998, Aug. 2014

2014
[49]

Mean-ﬁeld PHD ﬁlters based on g eneralized Feynman- Kac ﬂow,

M. Pace and P . Del Moral, “Mean-ﬁeld PHD ﬁlters based on g eneralized Feynman- Kac ﬂow,” IEEE J. Sel. Topics Sig. Proc. , vol. 7, no. 3, pp. 484–495, Jun. 2013

2013
[50]

M. S. Grewal and A. P . Andrews, Kalman ﬁltering: Theory and practice using MATLAB, 3rd ed., Wiley-IEEE Press, 2008

2008
[51]

An introductio n to Kalman ﬁltering with MA TLAB examples,

N. Kovvali, M. Banavar, and A. Spanias, “An introductio n to Kalman ﬁltering with MA TLAB examples,” Synthesis Lectures Sig. Proc. , Springer, 2013

2013
[52]

Pha se-based UHF RFID tracking with nonlinear Kalman ﬁltering and smoothing,

S. Sarkka, V . V . Viikari, M. Huusko and K. Jaakkola, “Pha se-based UHF RFID tracking with nonlinear Kalman ﬁltering and smoothing,” IEEE Sensors J. , vol. 12, no. 5, pp. 904–910, May 2012

2012
[53]

Discrete time q-lag maximum likelihood FIR smoothing and iterative recursive algorith m,

S. Zhao, J. Wang, Y . S. Shmaliy and F. Liu, “Discrete time q-lag maximum likelihood FIR smoothing and iterative recursive algorith m,” IEEE Trans. Sig. Proc. , vol. 69, pp. 6342–6354, 2021

2021
[54]

Spectral estimation via the hi gh-order Y ule-Walker equations,

Y . Chan and R. Langford, “Spectral estimation via the hi gh-order Y ule-Walker equations,” IEEE Trans. Acoust. Speech Sig. Proc. , vol. 30, no. 5, pp. 689–698, Oct. 1982

1982