Title : ( Deep Learning‐Driven Semantic Communication With Attention Modules )
Authors: Mohammad Ali Amirabadi ,Access to full-text not allowed by authors
Abstract
In this study, an innovative architecture is proposed to enhance the performance of semantic communication networks byleveraging deep learning and joint source-channel coding. A fundamental challenge in this field is the strong dependence ofconventional networks on a fixed signal-to-noise ratio (SNR) during training, which leads to performance degradation undervarying channel conditions. To address this limitation, we introduce a novel attention-based approach that enables dynamicadaptation to different SNR levels, ensuring more stable and optimized communication performance. The proposed model learnsmore generalized features that exhibit greater resilience to channel variations. To evaluate its effectiveness, extensive simulationswere conducted, comparing the performance of the proposed architecture with DeepSC, a state-of-the-art benchmark modelin the field. While the baseline model, trained at a single SNR, experiences performance drops under mismatched conditions,the proposed model, trained across a range of SNRs, achieves improvement of 16.2%, 30.8%, 42.8%, and 53.8% for 1, 2, 3, and4-gram precisions, respectively, in bilingual evaluation understudy score and an 11.4% increase in sentence similarity acrosschallenging low-SNR conditions. Furthermore, the model maintains robust performance with 48% less training data, highlightingits efficiency and data efficiency under practical constraints. These gains confirm the model’s superior adaptability and high-quality data reconstruction under diverse conditions. The results of this study underscore the significant benefits of attention-basedarchitectures in semantic communication, particularly in environments with unpredictable channel variations, and highlight theirpotential for reliable deployment in real-world applications.
Keywords
, Deep Learning, Semantic Communication, Attention Modules@article{paperid:1104604,
author = {Amirabadi, Mohammad Ali},
title = {Deep Learning‐Driven Semantic Communication With Attention Modules},
journal = {IET Communications},
year = {2025},
volume = {19},
number = {1},
month = {October},
issn = {1751-8628},
keywords = {Deep Learning; Semantic Communication; Attention Modules},
}
%0 Journal Article
%T Deep Learning‐Driven Semantic Communication With Attention Modules
%A Amirabadi, Mohammad Ali
%J IET Communications
%@ 1751-8628
%D 2025