EUVIMEDEUVIMED entdecken
Open Access · CC BY-NC-ND 4.0

Neural Network‐Based Successive Interference Cancellation for Non‐Orthogonal Multiple Access Downlink Transmission

Son‐Nam Jo, Hak‐Chol Ri, Chol Jong

IET Communications · 2026-01 · 20 (1)

DOI: 10.1049/cmu2.70233 ↗

Kurz gefasst

Fachartikel in der Zeitschrift IET Communications, online erschienen am 07.11.2026. Verfasst von Son‐Nam Jo und 2 weiteren Autor:innen. Die Arbeit nennt 54 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY-NC-ND 4.0).

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Non‐orthogonal multiple access (NOMA) is one of the most promising multiple access techniques for the fifth‐generation wireless communication systems to provide higher spectral efficiency and massive connectivity. NOMA enables multiple users to concurrently share the same resources by employing superposition coding at the transmitter and successive interference cancellation (SIC) at the receiver. However, SIC that requires accurate knowledge of both the channel model and channel state information (CSI) is challenged by the receiver complexity and error propagation. In this paper, we propose a neural network‐based SIC receiver referred to as N‐SIC, which learns to detect the transmitted symbol in a data‐driven manner in the downlink NOMA system without requiring any prior knowledge of the channel model and accurate CSI. Moreover, it can produce accurate soft outputs using the log‐likelihood ratio‐input forward error correction decoder for decoding each coded bit. In particular, we propose the federated learning method for N‐SIC to detect the desired signal under dynamic changes of the SIC order of NOMA users without the need for neural network reconstruction or retraining parameters. The simulation results demonstrate that under perfect CSI, the proposed N‐SIC approaches the performance of conventional SIC and outperforms the existing DL‐based SIC method. Furthermore, under imperfect CSI, N‐SIC shows an outstanding performance over two benchmark methods. Finally, we show that the federated learning‐based N‐SIC significantly achieves an improved performance compared with the existing DL‐based SIC receiver without the need for neural network reconstruction or retraining parameters with both perfect and imperfect CSI under dynamic changes of the SIC order of users.

Zitieren

APA

Jo, S., Ri, H., & Jong, C. (2026). Neural Network‐Based Successive Interference Cancellation for Non‐Orthogonal Multiple Access Downlink Transmission. IET Communications, 20(1). https://doi.org/10.1049/cmu2.70233

Vancouver

Jo S, Ri H, Jong C. Neural Network‐Based Successive Interference Cancellation for Non‐Orthogonal Multiple Access Downlink Transmission. IET Communications. 2026;20(1). doi:10.1049/cmu2.70233

BibTeX
@article{Jo2026,
  author = {Jo, Son‐Nam and Ri, Hak‐Chol and Jong, Chol},
  title = {Neural Network‐Based Successive Interference Cancellation for Non‐Orthogonal Multiple Access Downlink Transmission},
  journal = {IET Communications},
  year = {2026},
  volume = {20},
  number = {1},
  doi = {10.1049/cmu2.70233}
}
Infokarte | EUVIMED

Artikeldaten derzeit nicht verfügbar.

Recherche