Greedy and Transformer-Based Multi-Port Selection for Slow Fluid Antenna Multiple Access

Darian Perez-Adan; Jose P. Gonzalez-Coma; F. Javier Lopez-Martinez; Luis Castedo

cs.AIApr 6, 2026

Greedy and Transformer-Based Multi-Port Selection for Slow Fluid Antenna Multiple Access

Authors: Darian Perez-Adan, Jose P. Gonzalez-Coma, F. Javier Lopez-Martinez, Luis Castedo

Organizations: Department of Computer Engineering, University of A Coruña, CITIC, A Coruña, Spain · Defense University Center at the Spanish Naval Academy, Marín, Spain · Dept. Signal Theory, Networking and Communications, Research Centre for Information and Communication Technologies (CITIC-UGR), University of Granada, 18071, Granada, Spain

Abstract

We address the port-selection problem in fluid antenna multiple access (FAMA) systems with multi-port fluid antenna (FA) receivers. Existing methods either achieve near-optimal spectral efficiency (SE) at prohibitive computational cost or sacrifice significant performance for lower complexity. We propose two complementary strategies: (i) GFwd+S, a greedy forward-selection method with swap refinement that consistently outperforms state-of-the-art reference schemes in terms of SE, and (ii) a Transformer-based neural network trained via imitation learning followed by a Reinforce policy-gradient stage, which approaches GFwd+S performance at lower computational cost.

Figures & tables

Fig. 1: Transformer-based NN port selector architecture: 2 training phases.

Parameter	Value	Parameter	Value
$P$	100	$d$ (model dim.)	192
$L$	8	$h$ (heads)	6
$K=N_{\text{t}}$	10	$N_{\ell}$ (layers)	5
$W$	$4\lambda$	$d_{\mathrm{ff}}$	384
Correlation	Jakes [ 12 ]	Dropout	0.05

TABLE I: Simulation Parameters

Method	Complexity	Time (ms)
Slow-FAMA [ 4 ]	$\mathcal{O}(PK)$	0.28
DC [ 6 ]	$\mathcal{O}(PK+L^{3})$	0.42
CUMA [ 5 ]	$\mathcal{O}(P)$	0.15
GFwd (prop.)	$\mathcal{O}(PL^{4})$	90.90
GFwd+S (prop.)	$\mathcal{O}(RPL^{4})$	385.13
GEPort [ 6 ]	$\mathcal{O}\big((P{-}L)P^{3}\big)$	232.04

TABLE II: Computational Complexity and Inference Times

Fig. 2: Validation SE during training at SNR

\in\{10,15,20\}

dB.

Fig. 3: Average SE versus SNR for

P=100

L=8

, and

K=10

Fig. 4: Average SE versus swap rounds

R

for

P=100

L=8

, and

K=10

. Dashed lines indicate GEPort reference performance at each SNR .

Fig. 5: Average SE vs.

K

(users) for

P=100

L=8

, and

\mathrm{SNR}=15

dB.

Fig. 6: Average SE vs. RF chains

L

for

P=100

K=10

\mathrm{SNR}=15

dB.

Fig. 7: Average SE versus total ports

P

for

K=10

L=8

\mathrm{SNR}=15

dB.

Wireless Communications

Wireless communications studies how information is sent over radio links and how networks manage signals, spectrum, and resources reliably.

See all 399 papers

Beyond LLM-GA: Secure Fluid Antenna Systems with ReEvo-Designed Memetic Algorithm
Oct 7, 2026Hanyong Xu, Zhaolai Dang, Tong ZhangWireless Communications Metaheuristic Optimization
School of Information, Harbin Institute of Technology, Shenzhen, Shenzhen 518055, China · Guangdong Provincial Key Laboratory of Aerospace Communication and Networking Technology, Harbin Institute of Technology, Shenzhen, 518055, China
AirPASS: Over-the-Air Federated Learning via Pinching Antenna Systems
Jul 7, 2026Seyed Mohammad Azimi-Abarghouyi, Christopher G. BrintonFederated Learning Aggregation Wireless Communications
Department of Electrical Engineering, Chalmers University of Technology, Gothenburg, Sweden · School of Electrical and Computer Engineering, Purdue University, West Lafayette, IN USA
Low-Altitude Fluid Antenna Network with Multi-Agent Reinforcement Learning
Aug 28, 2026Tong Zhang, Yanfei Su, Shuai Wang +3Wireless Communications Wireless Resource Allocation
Guangdong Provincial Key Laboratory of Aerospace Communication and Networking Technology, Harbin Institute of Technology, Shenzhen, 518055, China · Harbin Institute of Technology, Shenzhen, China · Shenzhen Institutes of Advanced Technology, Chinese Academy of Sciences, Shenzhen 518055, China +3

Oct 7, 2026J/K move · Enter open · S save

Beyond LLM-GA: Secure Fluid Antenna Systems with ReEvo-Designed Memetic Algorithm

Hanyong Xu, Zhaolai Dang, Tong Zhang

School of Information, Harbin Institute of Technology, Shenzhen, Shenzhen 518055, China · Guangdong Provincial Key Laboratory of Aerospace Communication and Networking Technology, Harbin Institute of Technology, Shenzhen, 518055, China

Wireless Communications Metaheuristic Optimization Large Language Model-Guided Optimization

Open paper AskPDF arXiv

Fluid antenna systems (FASs) offer significant spatial flexibility, yet securing them against eavesdropping is critical for practical FAS deployment in military, satellite, and internet-of-things networks. Although large language model (LLM)-assisted genetic algorithms (LLM-GAs) can address this secure FAS port selection problem, whether further algorithmic improvement is possible warrants deeper investigation. To this end, we propose a memetic algorithm based on reflective evolution (ReEvo). Unlike the state-of-the-art LLM-GAs, which design only crossover or mutation operators with an LLM, our algorithm leverages an LLM to evolve dedicated crossover, mutation, and local-search operators offline. These operators are then embedded into a memetic search framework, thereby obviating any online LLM queries during execution. Simulation results at equal generation counts demonstrate that our proposed algorithm achieves a higher secure sum-rate than the conventional GA and the state-of-the-art LLM-GAs.

Abstract

Wireless Communications

Explore similar work

Beyond LLM-GA: Secure Fluid Antenna Systems with ReEvo-Designed Memetic Algorithm

AirPASS: Over-the-Air Federated Learning via Pinching Antenna Systems

Low-Altitude Fluid Antenna Network with Multi-Agent Reinforcement Learning