Source-linked AI summary
An Information-Theoretic Characterization of MIMO-FAS: Optimization, Diversity-Multiplexing Tradeoff and $q$-Outage Capacity
Wee Kiat New, Kai-Kit Wong, Hao Xu, Kin-Fai Tong, Chan-Byoung Chae
TL;DR
The paper studies how multi-port, two-dimensional fluid antennas can provide additional degrees of freedom beyond conventional MIMO for 6G-oriented communication systems. It proposes QR MIMO-FAS and derives DMT and q-outage capacity, reporting lower outage and higher achievable rates than traditional MIMO and MIMO-AS under the studied settings.
Problem
The paper investigates whether multi-port fluid antennas on 2D surfaces can provide additional spatial diversity and performance beyond traditional MIMO antenna-selection systems.
Method
The paper proposes QR MIMO-FAS, combining joint port selection, transmit and receive beamforming, and power allocation, then analyzes DMT and q-outage capacity.
Results
QR MIMO-FAS achieves similar high-SNR average rate to optimal MIMO-FAS, reaches outage probabilities on the order of 10^-3, and outperforms traditional MIMO and MIMO-AS in rate and outage.
Takeaways & Limitations
MIMO-FAS can provide substantially greater diversity and q-outage capacity than traditional MIMO and MIMO-AS under the paper's modeled conditions.
Abstract
from arXiv · showhide
Multiple-input multiple-output (MIMO) system has been the defining mobile communications technology in recent generations. With the ever-increasing demands looming towards the sixth generation (6G), we are in need of additional degrees of freedom that deliver further gains beyond MIMO. To this goal, fluid antenna system (FAS) has emerged as a new way to obtain spatial diversity using reconfigurable position-switchable antennas. Considering the case with more than one ports activated on a 2D fluid antenna surface at both ends, we take the information-theoretic approach to study the achievable performance limits of the MIMO-FAS. First of all, we propose a suboptimal scheme, referred to as QR MIMO-FAS, to maximize the rate at high signal-to-noise ratio (SNR) via joint port selection, transmit and receive beamforming and power allocation. We then derive the optimal diversity and multiplexing tradeoff (DMT) of MIMO-FAS. From the DMT, we highlight that MIMO-FAS outperforms traditional MIMO antenna systems. Further, we introduce a new metric, namely q-outage capacity, which can jointly consider rate and outage probability. Through this metric, our results indicate that MIMO-FAS surpasses traditional MIMO greatly.
I. INTRODUCTION
The introduction motivates MIMO-FAS as a way to add spatial degrees of freedom beyond MIMO, while noting that its multi-port performance and optimal DMT remain insufficiently understood.
- 6G motivates new technologies that can exceed the performance of current MIMO systems with the same bandwidth.
- FAS uses software-controllable structures or switchable positions to reconfigure antenna characteristics and access different spatial channels.
- Closely spaced ports can be strongly correlated, making spatial correlation an important factor in FAS performance analysis.
- Prior work reported capacity and outage improvements for FAS, but accurate correlation models can produce an outage floor and the optimal MIMO-FAS DMT was unknown.
- MIMO-FAS activates multiple ports on 2D surfaces at both transmitter and receiver, with dynamically adjustable radiating-element positions and potentially many preset locations.
B. Contributions
The paper develops a correlation-aware MIMO-FAS framework, rate-optimization schemes, an optimal DMT characterization, and q-outage-capacity comparisons with traditional systems.
- The paper models 2D MIMO-FAS at both ends while incorporating spatial correlation in a 3D rich-scattering environment.
- It formulates rate maximization through joint port selection, transmit and receive beamforming, and power allocation.
- QR MIMO-FAS uses suboptimal port selection, beamforming, and power allocation to achieve polynomial time complexity at high SNR.
- The paper derives an outer DMT bound and combines it with QR MIMO-FAS to obtain the optimal DMT of MIMO-FAS.
- MIMO-FAS has greater diversity gain than MIMO and MIMO antenna selection when the total number of active ports or antennas is the same.
- q-outage capacity jointly considers rate and outage probability, and MIMO-FAS outperforms traditional MIMO and antenna selection under this metric.
C. Organization and Notations
The paper is organized around the system model, QR MIMO-FAS, DMT analysis, numerical comparisons, and conclusion, with notation conventions and a key-notation table provided.
- Organization: Sections II–VI cover the system model, QR MIMO-FAS, optimal DMT analysis, numerical comparisons, and the conclusion.
- Notations: Scalar, vector, and matrix variables use distinct typographic conventions, while transpose, conjugate transpose, determinant, rank, and trace are defined.
- Notations: The notation section also defines norms, logarithms, expectations, Kronecker products, basis vectors, diagonal matrices, and pseudoinverses.
- Notations: Table I lists the meanings of key notations used throughout the mathematical development.
II. SYSTEM MODEL
The system model represents point-to-point MIMO-FAS with uniformly distributed ports on 2D surfaces, selectable active ports, beamforming, power allocation, and spatially correlated channels.
- The transmitter and receiver each use a 2D fluid antenna surface of area W_s with N_s uniformly distributed ports, while only n_s ports can be active.
- Port locations are represented on a grid, with a mapping function converting 2D port coordinates into scalar port labels.
- The 3D rich-scattering model uses spherical-Bessel-function spatial correlations between ports.
- The channel covariance incorporates transmit- and receive-side spatial correlation through a Kronecker product.
- Activation matrices select distinct transmit and receive ports, while beamforming matrices satisfy a unit-norm constraint.
- The MIMO-FAS rate depends on the effective channel and power-allocation matrix, whose trace is constrained by the transmit SNR.
- Port spatial correlation can allow full channel-state information to be obtained from only a small number of observed ports or training samples.
III. QR MIMO-FAS: SUBOPTIMAL PORT SELECTION, BEAMFORMING AND POWER ALLOCATION
The paper formulates rate maximization as a coupled port-selection, beamforming, and power-allocation problem, then proposes QR MIMO-FAS as a lower-complexity high-SNR solution.
- The original rate-maximization problem is non-convex because its variables are mutually coupled and its domain is non-convex.
- The global optimum requires exhaustive search, SVD, and waterfilling, but has non-polynomial complexity O(N_tx^n_tx N_rx^n_rx).
- QR MIMO-FAS decouples optimization into port selection and beamforming with power allocation, using high-SNR rate dependence on det(H̄H̄^H).
- Strong rank-revealing QR factorization selects ports by permuting columns according to the largest Ω_k,l, yielding a suboptimal solution.
- Given selected ports, SVD and waterfilling solve the optimal beamforming and power-allocation subproblem.
- The proposed scheme has polynomial time complexity and significantly reduces computation relative to the global optimum.
IV. OPTIMAL DMT
The paper derives the optimal DMT of MIMO-FAS by reducing spatially correlated channels to finite-rank representations and matching QR MIMO-FAS to the resulting outer bound.
- The optimal DMT of a full-rank correlated channel equals that of an equivalent channel G.
- Using only n_rx×n_tx channels yields a piecewise linear DMT connecting (n_min, 0) and {r, (N_rx−r)(N_tx−r)}.
- For grid-placed antennas at least half a wavelength apart with full-rank correlation matrices, antenna selection has a piecewise linear DMT connecting (n_min, 0) and {r, (w_rx−r)(w_tx−r)}.
- For finite W_rx and W_tx, the MIMO-FAS DMT has a piecewise linear outer bound derived through singular-value interlacing and outage analysis.
- Spatially fully correlated channels prevent conventional joint-PDF simplifications from retaining tractable singular-value exponents.
- The DMT of QR MIMO-FAS equals the outer bound and is therefore the optimal DMT of MIMO-FAS.
R (SNR)
The paper studies how the effective rank of the spatial-correlation matrix can be estimated and used to characterize MIMO-FAS performance as antenna-port counts grow.
- The proposed methods estimate N′_s from N_s1, N_s2, W_s1, and W_s2 and provide certificates for reducing or reconstructing J_s.
- Table II reports estimates of N′_s for different surface dimensions and port configurations.
- MIMO-FAS yields massive diversity gains when the multiplexing gain satisfies r < n_min.
- The q-outage capacity is introduced to jointly evaluate reliably transmitted rate and outage probability for a fixed target rate q independent of SNR.
MIMO-FAS (SNR, q)
The paper states that MIMO-FAS provides benefits over traditional antenna systems.
- MIMO-FAS can harness benefits over a traditional antenna system using (47).
V. RESULTS AND DISCUSSIONS
The results evaluate QR MIMO-FAS and benchmark schemes across port layouts, SNR, surface size, mutual coupling, outage targets, and q-outage capacity. QR MIMO-FAS generally benefits from port selection and spatial diversity, while performance depends on SNR, surface dimensions, and port resolution.
- 2D port placement: 2D port distributions achieve much lower outage probability than 1D distributions with the same number of ports.The additional spatial dimension provides more spatial diversity.
- Average rate: At high SNR, QR MIMO-FAS achieves a similar average rate to optimal MIMO-FAS and exceeds traditional MIMO when the number of active ports or antennas is equal.Its average rate scales like n_s log SNR for n_s from 1 to 6, then exhibits diminishing gains from 7 to 12.
- Mutual coupling: QR MIMO-FAS and random MIMO-FAS remain broadly similar with and without mutual coupling, whereas greedy MIMO-FAS degrades more as N_s increases.The comparison considers liquid-based and RF pixel-based fluid antennas.
- Large surface size: For large N_s, QR MIMO-FAS achieves 1bps/Hz higher average rate than MIMO-AS and 6bps/Hz higher than MIMO, with outage probability on the order of 10^-3.MIMO and random or greedy MIMO-FAS have outage probabilities near 0.99 in the reported setting, while MIMO-AS is on the order of 10^-2.
- Surface dimensions: Increasing W_s decreases QR MIMO-FAS outage probability without bound when N_s is sufficiently large, while MIMO remains near-one outage and MIMO-AS declines more slowly.Average rates increase and then plateau, and the results suggest determining N_s according to W_s.
- Diversity and outage: For q below n_min log SNR, QR MIMO-FAS outage probability falls from about 10^-1 to 10^-5 within 6bits/Hz, outperforming MIMO-AS and MIMO.The optimal DMT is also the optimal DMT of MIMO-FAS; its maximum diversity is approximately 529 versus 81 for MIMO-AS and 16 for 4×4 MIMO when W_rx = W_tx = 1λ^2.
VI. CONCLUSIONS
The paper characterizes MIMO-FAS performance limits with a spatially correlated 2D surface model, proposes QR MIMO-FAS, and derives its optimal DMT. Results show near-optimal high-SNR rate, scalable rate and outage behavior, and massive diversity gains over traditional systems.
- The proposed QR MIMO-FAS jointly selects ports, performs transmit and receive beamforming, and allocates power to maximize high-SNR rate.The paper also derives the outer bound and optimal DMT, establishing MIMO-FAS performance limits.
- QR MIMO-FAS achieves a similar rate to optimal MIMO-FAS in the high-SNR regime.
- As N_s increases, QR MIMO-FAS average rate and outage probability approach limits when other MIMO-FAS parameters are fixed.
- As W_s increases, QR MIMO-FAS average rate improves up to a certain level, while outage probability decreases without bound.
- For the same multiplexing gain, MIMO-FAS achieves massive diversity gain compared with traditional MIMO and MIMO-AS systems.
APPENDIX I: SPATIAL CORRELATION OF 2D FLUID ANTENNA SURFACE OVER 3D SCATTERING ENVIRONMENT
The appendix develops the spatial-correlation model for a 2D fluid-antenna surface in a 3D scattering environment and discusses mutual-coupling representations and assumptions.
- Spatial correlation: The 2D fluid-antenna surface is modeled through port positions, array responses, spatial-correlation matrices, and spherical-wave expansions.The formulation uses spherical Bessel functions, spherical harmonics, and a 3D isotropic scattering assumption.
- Spatial correlation: The spatial-correlation matrix can be reduced to a full-rank symmetric matrix by removing dependent rows and columns.The reduction identifies the effective full rank of the original correlation structure.
- Mutual coupling: The liquid-based fluid-antenna model includes mutual coupling only between active ports, with return-loss and isolation levels set to −15dB and 30dB.The mutual-coupling matrix depends on antenna, load, and mutual impedances; active ports are modeled as dipoles of length 0.5λ and width 0.001λ.
- Mutual coupling: The RF pixel-based model allows mutual coupling whether pixels are on or off, so antenna design must improve the S-matrix.For widely separated active ports, the mutual-coupling matrix is approximated by the identity.
- Modeling scope: The evaluated mutual-coupling model is conservative because optimization omits coupling effects, while joint optimization could further improve MIMO-FAS performance.The stated scope focuses on performance without mutual coupling when selecting transmit and receive beamformers.