EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

GMCANet: A Graph MinCut-Clustering Attention Network for Automatic Modulation Classification

Haonan Liu, Wenzhe Fan, Jun Xie

IEEE Access · 2026 · 14 · 47686-47699

DOI: 10.1109/access.2026.3676832 ↗

Kurz gefasst

Fachartikel in der Zeitschrift IEEE Access, 2026. Verfasst von Haonan Liu und 2 weiteren Autor:innen. Die Arbeit nennt 32 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: National Natural Science Foundation of China.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

In intelligent spectrum monitoring, due to the constraints of dynamic electromagnetic environments, balancing the trade-off between computational complexity and recognition accuracy remains a bottleneck for Automatic Modulation Classification (AMC). To tackle this problem, a method called Graph MinCut-Clustering Attention Network (GMCANet) is proposed, which offers a lightweight yet effective solution for deployment. In GMCANet, the raw I/Q signals are first processed via data augmentation and feature extraction to generate informative node features. Next, an adaptive learning mechanism is employed to learn a dynamic adjacency matrix, capturing the underlying relationships among signals. Finally, to coarsen the graph structure and extract representative features, a Graph MinCut-Clustering Attention (GMCA) block is introduced by leveraging spectral clustering and dynamic attention mechanisms, where adjacency learning and clustering are jointly optimized within an end-to-end training framework. Experimental results on the RadioML2016 dataset show that GMCANet achieves an average accuracy of 63.95%, achieving a competitive accuracy–complexity trade-off compared with representative baselines.

Förderung

  • National Natural Science Foundation of China

Zitieren

APA

Liu, H., Fan, W., & Xie, J. (2026). GMCANet: A Graph MinCut-Clustering Attention Network for Automatic Modulation Classification. IEEE Access, 14, 47686-47699. https://doi.org/10.1109/access.2026.3676832

Vancouver

Liu H, Fan W, Xie J. GMCANet: A Graph MinCut-Clustering Attention Network for Automatic Modulation Classification. IEEE Access. 2026;14:47686-47699. doi:10.1109/access.2026.3676832

BibTeX
@article{Liu2026,
  author = {Liu, Haonan and Fan, Wenzhe and Xie, Jun},
  title = {GMCANet: A Graph MinCut-Clustering Attention Network for Automatic Modulation Classification},
  journal = {IEEE Access},
  year = {2026},
  volume = {14},
  pages = {47686--47699},
  doi = {10.1109/access.2026.3676832}
}
Kostenlos starten →