@article{M42F8B1A0,
title = "Automatic Modulation Recognition Using Data-Efficient image Transformers(DeiT)",
journal = "The Journal of Korean Institute of Communications and Information Sciences",
year = "2025",
issn = "1226-4717",
doi = "10.7840/kics.2025.50.2.245",
author = "Youjeong Yoon, Hae Chung, Wansu Lim",
keywords = "Automatic modulation recognition, 6G, Data-efficient Image Transformers(Deit), Wireless communication systems",
abstract = "Automatic Modulation Recognition (AMR) is a technology that automatically identifies the modulation of a received signal in a wireless communication system, which plays an important role in detecting jamming signals and improves the performance of military and commercial communication systems. With the recent advances in deep learning, AMR has been actively researched by introducing deep learning techniques in the field of AMR.
In this paper, we propose a new algorithm to automatically recognize modulation schemes using the Data-efficient Image Transformers (DeiT) model, which can efficiently achieve high performance without the need for large datasets. The DeiT model utilizes knowledge distillation techniques to combine the advantages of each model by using the structure of Vision Transformer (ViT) while retaining the inductive bias of CNN architecture.
Experimental results show that AMR based on DeiT is on average 8.2% more accurate than modulation recognition using ViT."
}