EUVIMEDEUVIMED entdecken
Crossref

Diffusion Expectation-Maximization for Joint MU-MIMO Channel Estimation and Signal Detection in Fluid Antenna Systems

Sihan Du, Chao Wang, Haibin Zhang

IEEE Transactions on Cognitive Communications and Networking · 2026 · 12 · 9583-9599

DOI: 10.1109/tccn.2026.3706698 ↗

Kurz gefasst

Fachartikel in der Zeitschrift IEEE Transactions on Cognitive Communications and Networking, 2026. Verfasst von Sihan Du und 2 weiteren Autor:innen. Die Arbeit nennt 48 Literaturangaben. Gefördert von: National Natural Science Foundation of China, Funds for International Cooperation and Exchange of the National Natural Science Foundation of China, Key Research and Development Program of Shaanxi.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

This paper focuses on MU-MIMO (multiuser multiple-input multiple-output) channel estimation and signal detection in uplink fluid antenna systems (FASs). On one hand, FASs incorporate a large number of nearly continuously reconfigurable ports, but the number of radio frequency (RF) chains available for real-time channel measurements is extremely limited, requiring high-precision channel reconstruction with limited pilots. On the other hand, the conventional separate processing framework of “channel estimation first, signal detection later” has inherent limitations in FASs. Channel estimation errors directly propagate to the signal detection phase, degrading detection performance below its theoretical upper bound, and channel state information (CSI) acquisition requires continuous port switching, which easily leads to severe detection performance degradation due to channel time variation. To address these challenges, we propose a joint processing paradigm based on diffusion expectation-maximization (EM) for FASs, which utilizes pilots from a few ports to simultaneously realize channel estimation and signal detection. Specifically, in the expectation step (E-step), leveraging the powerful high-dimensional data modeling and generation capabilities of diffusion models (DMs), we construct the channel prior distribution in the training phase, and then generate the posterior channel samples to approximate the expected log-likelihood in the inference phase. In the maximization step (M-step), we solve the maximum a posteriori (MAP) estimation problem for FAS signal detection by using a plug-and-play (PnP) denoising prior method. Then, the E-step and M-step are performed alternately to realize the joint estimation. Furthermore, since the DMs are long to run, we propose a fast joint processing paradigm for FASs. Simulation results demonstrate that under limited pilots, the proposed joint processing paradigm achieves higher accuracy in both channel estimation and signal detection compared with the separate processing paradigm in FASs.

Förderung

  • National Natural Science Foundation of China
  • Funds for International Cooperation and Exchange of the National Natural Science Foundation of China
  • Key Research and Development Program of Shaanxi
  • Natural Science Basis Research Plan in Shaanxi Province of China
  • National Science Fund for Distinguished Young Scholars of China

Zitieren

APA

Du, S., Wang, C., & Zhang, H. (2026). Diffusion Expectation-Maximization for Joint MU-MIMO Channel Estimation and Signal Detection in Fluid Antenna Systems. IEEE Transactions on Cognitive Communications and Networking, 12, 9583-9599. https://doi.org/10.1109/tccn.2026.3706698

Vancouver

Du S, Wang C, Zhang H. Diffusion Expectation-Maximization for Joint MU-MIMO Channel Estimation and Signal Detection in Fluid Antenna Systems. IEEE Transactions on Cognitive Communications and Networking. 2026;12:9583-9599. doi:10.1109/tccn.2026.3706698

BibTeX
@article{Du2026,
  author = {Du, Sihan and Wang, Chao and Zhang, Haibin},
  title = {Diffusion Expectation-Maximization for Joint MU-MIMO Channel Estimation and Signal Detection in Fluid Antenna Systems},
  journal = {IEEE Transactions on Cognitive Communications and Networking},
  year = {2026},
  volume = {12},
  pages = {9583--9599},
  doi = {10.1109/tccn.2026.3706698}
}
Kostenlos starten →