A Modified Kolmogorov–Arnold Network for Automatic Modulation Classification
IEEE Photonics Technology Letters · 01.08.2026 · 38 (15) · 1130-1133
DOI: 10.1109/lpt.2026.3677693 ↗Kurz gefasst
Fachartikel in der Zeitschrift IEEE Photonics Technology Letters, erschienen am 01.08.2026. Verfasst von Kaige Yang und 7 weiteren Autor:innen. Die Arbeit wurde bisher 1-mal zitiert (Crossref) und nennt 16 Literaturangaben. Gefördert von: National Natural Science Foundation of China, National Key Research and Development Program of China, Natural Science Foundation of Shanghai.
Abstract (Originalsprache)
Automatic modulation classification (AMC) is a fundamental signal-awareness mechanism in intelligent communication systems, including cognitive radio, dynamic spectrum sensing, electronic warfare, and signal intelligence, enabling adaptive transmission, resource allocation, and receiver optimization. However, under atmospheric turbulence and nonlinear channel impairments, conventional deep-learning-based AMC, particularly those based on multilayer perceptrons, suffers from limited interpretability, slow convergence, and degraded robustness. Kolmogorov-Arnold Networks (KANs) provide a lightweight and explainable alternative; nevertheless, their generic activation function libraries are not aligned with turbulence-induced statistical characteristics, leading to suboptimal gradient propagation and feature discrimination. To address this mismatch, we derive a closed-form expression for the joint atmospheric-receiver noise model and construct a distribution-aware activation function library based on its Fox-H function representation. Experimental results demonstrate that the modified KAN achieves higher classification accuracy and reduced computational complexity in atmospheric turbulence channels by tailoring the activation functions to the channel noise statistics.
Förderung
- National Natural Science Foundation of China
- National Key Research and Development Program of China
- Natural Science Foundation of Shanghai
- Shanghai Collaborative Innovation Project
- High Performance Computing Center, University of Shanghai for Science and Technology
Zitieren
APA
Yang, K., Yan, Y., Hou, Y., Li, Y., Zhang, L., Xu, B., Wang, K., & Zhang, D. (2026). A Modified Kolmogorov–Arnold Network for Automatic Modulation Classification. IEEE Photonics Technology Letters, 38(15), 1130-1133. https://doi.org/10.1109/lpt.2026.3677693
Vancouver
Yang K, Yan Y, Hou Y, Li Y, Zhang L, Xu B, et al. A Modified Kolmogorov–Arnold Network for Automatic Modulation Classification. IEEE Photonics Technology Letters. 2026;38(15):1130-1133. doi:10.1109/lpt.2026.3677693
BibTeX
@article{Yang2026,
author = {Yang, Kaige and Yan, Yu and Hou, Yaxi and Li, Yishan and Zhang, Leihong and Xu, Banglian and Wang, Kaimin and Zhang, Dawei},
title = {A Modified Kolmogorov–Arnold Network for Automatic Modulation Classification},
journal = {IEEE Photonics Technology Letters},
year = {2026},
volume = {38},
number = {15},
pages = {1130--1133},
doi = {10.1109/lpt.2026.3677693}
}
EUVIMED