Technical Summary: MISO Downlink with Fluid Antenna Multiple Access

1. Problem Statement

Fluid Antenna Multiple Access (FAMA) is an emerging paradigm where users dynamically reposition a single radio-frequency (RF) chain antenna within a compact aperture to select the port offering the strongest instantaneous signal. While FAMA offers micro-scale spatial diversity, its performance in multiuser downlink systems remains analytically uncharacterized when combined with linear precoding and spatial correlation.

Existing literature primarily focuses on single-input single-output (SISO) configurations, independent ports, or strongest-port statistics without accounting for multiuser precoding or spatial correlation. Consequently, the joint effects of Maximum Ratio Transmission (MRT) and Zero-Forcing (ZF) beamforming, interference, port-wise Signal-to-Interference Ratio (SIR) statistics, and spatial correlation on outage performance have not been derived in closed form. This paper addresses the gap by developing a unified analytical framework for the Multiple-Input Single-Output (MISO) downlink with FAMA users, specifically analyzing how linear precoding interacts with correlated port selection.

2. Methodology

The authors propose a unified analytical framework for a downlink MISO system where a Base Station (BS) with $M$ antennas serves $U$ single-RF users. Each user is equipped with a fluid antenna capable of switching among $N$ ports distributed over a linear aperture of length $W\lambda$.

System Model and Assumptions:

• Channel Model: The system operates under Rayleigh fading with an isotropic scattering environment. Spatial correlation between ports is modeled using a Bessel-type correlation function ($J_0$), where the channel at port $k$ is expressed as a correlated version of a reference port (Port 1) plus an independent innovation component.
• CSI Acquisition: The BS acquires instantaneous Channel State Information (CSI) only for the reference port (Port 1) of each user. Precoders (MRT or ZF) are designed solely based on this reference CSI.
• Precoding Strategies:
  • MRT: Beamformers align with the reference channel vector to maximize signal power.
  • ZF: Beamformers are designed to null interference at the reference ports.
• Performance Metric: The analysis focuses on the Signal-to-Interference Ratio (SIR), assuming an interference-limited regime where thermal noise is negligible compared to multiuser interference. The user selects the port $k$ that maximizes the instantaneous SIR.

Analytical Approach:

1. Statistical Characterization: The authors derive the exact probability distributions of the per-port SIR for both MRT and ZF precoding.
2. Correlation Modeling: A closed-form approximation for the cross-port SIR correlation coefficient is derived, accounting for the geometric overlap of ports and the effective array dimension.
3. Outage Analysis: Rigorous upper and lower bounds on the FAMA outage probability are established for arbitrary correlation levels, avoiding the intractability of the exact joint distribution.
4. Asymptotic Analysis: The paper investigates the scaling laws for large numbers of ports ($N$), large antenna arrays ($M$), and extreme SIR thresholds (small and large $\gamma$).

3. Key Contributions

A. Exact Per-Port SIR Distributions

The paper demonstrates that under both MRT and ZF precoding, the per-port SIR follows a Beta-prime distribution. The distribution is characterized by parameters $(a, b) = (M_{\text{eff}}, L)$, where $L = U-1$ is the number of interfering streams, and $M_{\text{eff}}$ is the effective signal dimension:

• MRT: $M_{\text{eff}} = M$.
• ZF: $M_{\text{eff}} = M - U + 1$.
  This distinction reveals that ZF reduces the effective array gain due to the degrees of freedom consumed by interference nulling.

B. Closed-Form Finite-Sum CDFs

The authors convert the incomplete Beta function representations of the SIR Cumulative Distribution Functions (CDFs) into closed-form finite-sum expressions. These formulas enable exact, tractable outage analysis without numerical integration for integer parameters.

C. Analytical Cross-Port SIR Correlation Model

This work provides the first closed-form expression for the correlation between SIRs across fluid-antenna ports. The derived correlation coefficient depends jointly on:

1. Geometric overlap parameters ($\mu_k \mu_\ell$).
2. The effective BS array dimension ($M_{\text{eff}}$).
   The analysis shows that MRT induces weaker cross-port correlation than ZF for the same geometry, as the larger $M_{\text{eff}}$ in MRT enhances the mixing effect across ports.

D. Rigorous Outage Bounds

Tight upper and lower bounds on the FAMA outage probability are derived for arbitrary cross-port correlation:

• Upper Bound: Corresponds to the single-port outage (fully correlated ports).
• Lower Bound: Corresponds to the independent-port case (ideal selection diversity).
  These bounds bracket the exact performance and become exact in the limiting cases of fully correlated and independent ports, respectively.

E. Asymptotic Scaling Laws

• Small-Threshold Regime: The single-port outage probability decays as $\gamma^{M_{\text{eff}}}$. Under FAMA port selection in a weakly correlated regime, the effective diversity order scales as $M_{\text{eff}}N$.
• Large-SIR Regime: The tail behavior is universally controlled by the number of interfering users ($L$), decaying as $\gamma^{-L}$, regardless of the precoding scheme.

4. Results and Numerical Validation

Numerical results based on Monte Carlo simulations validate the analytical framework:

• Distribution Accuracy: The finite-sum CDFs match simulation results exactly for both MRT and ZF.
• Correlation Model: The derived correlation expressions accurately capture the decay profile of SIR correlation across ports for various aperture sizes ($W$). MRT consistently exhibits lower correlation than ZF.
• Outage Bounds: The proposed bounds tightly bracket the simulated outage probability. In large apertures (weak correlation), the performance approaches the independent-port benchmark; in small apertures (strong correlation), it approaches the single-port limit.
• Precoder Comparison:
  • MRT: Achieves stronger selection gains and weaker port correlation, making it preferable when the BS has ample spatial degrees of freedom ($M \gg U$) and in moderate interference regimes.
  • ZF: Provides aggressive interference suppression but suffers from higher cross-port correlation and reduced effective diversity. It is advantageous in dense multiuser scenarios where interference suppression is critical, provided the aperture is large enough to mitigate the correlation penalty.

5. Significance and Claims

The paper claims to provide the first analytical characterization of per-port SIR statistics, cross-port correlation, and outage bounds for multiuser MISO downlink systems with FAMA under linear precoding.

Key significance points include:

• Unified Framework: It bridges the gap between fluid antenna port selection and multiuser precoding, offering explicit guidelines for port configuration and precoder selection.
• Design Insights: The results offer direct engineering insights, such as the trade-off between aperture size and selection gain, and the conditions under which MRT or ZF is preferable.
• Theoretical Benchmark: The derived bounds and distributions serve as theoretical performance benchmarks, isolating signal-processing effects from hardware imperfections.
• Limitations and Scope: The authors explicitly state that the analysis assumes idealized channel conditions (Rayleigh fading, perfect CSI at the reference port) and does not model mechanical actuation inaccuracies. The results are interpreted as optimistic upper bounds on performance, with future work suggested for Rician fading and imperfect CSI scenarios.

In summary, the paper establishes that while FAMA offers significant diversity gains, these gains are fundamentally constrained by spatial correlation and the specific linear precoding strategy employed, with MRT generally offering superior robustness and selection diversity compared to ZF in the analyzed MISO setting.