Guardado en:
Detalles Bibliográficos
Autores principales: Tardy, Xavier, Lefebvre, Grégoire, Kountouris, Apostolos, Fares, Haïfa, Nafkha, Amor
Formato: Preprint
Publicado: 2026
Materias:
Acceso en línea:https://arxiv.org/abs/2602.04728
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
_version_ 1866908940463243264
author Tardy, Xavier
Lefebvre, Grégoire
Kountouris, Apostolos
Fares, Haïfa
Nafkha, Amor
author_facet Tardy, Xavier
Lefebvre, Grégoire
Kountouris, Apostolos
Fares, Haïfa
Nafkha, Amor
contents We propose a cross-attention Transformer for joint decoding of uplink OFDM signals received by multiple coordinated access points. A shared per-receiver encoder learns the time-frequency structure of each grid, and a token-wise cross-attention module fuses the receivers to produce soft log-likelihood ratios for a standard channel decoder without explicit channel estimates. Trained with a bit-metric objective, the model adapts its fusion to per-receiver reliability and remains robust under degraded links, strong frequency selectivity, and sparse pilots. Over realistic Wi-Fi channels, it outperforms classical pipelines and strong neural baselines, often matching or surpassing a local perfect-CSI reference while remaining compact and computationally efficient on commodity hardware, making it suitable for next-generation coordinated Wi-Fi receivers.
format Preprint
id arxiv_https___arxiv_org_abs_2602_04728
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Scalable Cross-Attention Transformer for Cooperative Multi-AP OFDM Uplink Reception
Tardy, Xavier
Lefebvre, Grégoire
Kountouris, Apostolos
Fares, Haïfa
Nafkha, Amor
Signal Processing
Information Theory
Machine Learning
94A12, 68T05
I.2.6; C.2.1
We propose a cross-attention Transformer for joint decoding of uplink OFDM signals received by multiple coordinated access points. A shared per-receiver encoder learns the time-frequency structure of each grid, and a token-wise cross-attention module fuses the receivers to produce soft log-likelihood ratios for a standard channel decoder without explicit channel estimates. Trained with a bit-metric objective, the model adapts its fusion to per-receiver reliability and remains robust under degraded links, strong frequency selectivity, and sparse pilots. Over realistic Wi-Fi channels, it outperforms classical pipelines and strong neural baselines, often matching or surpassing a local perfect-CSI reference while remaining compact and computationally efficient on commodity hardware, making it suitable for next-generation coordinated Wi-Fi receivers.
title Scalable Cross-Attention Transformer for Cooperative Multi-AP OFDM Uplink Reception
topic Signal Processing
Information Theory
Machine Learning
94A12, 68T05
I.2.6; C.2.1
url https://arxiv.org/abs/2602.04728