EUVIMEDEUVIMED entdecken
Crossref

Uncertainty-Aware Multi-Modal Semantic Communication Over 6G Fading Channels with Adaptive Latent Correction

Muhammad Afzal Shah, Yang Tiemei, Kashif Bashir, Muhammad Suleman Soomro

International Journal of Innovative Science and Research Technology · 25.06.2026 · 1254

DOI: 10.38124/ijisrt/26jun260 ↗

Kurz gefasst

Fachartikel in der Zeitschrift International Journal of Innovative Science and Research Technology, online erschienen am 25.06.2026. Verfasst von Muhammad Afzal Shah und 3 weiteren Autor:innen. Die Arbeit nennt 40 Literaturangaben.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

The proposed Uncertainty-Aware Multi-Modal Semantic Communication (UAMM-SC) architecture serves as a fundamental concept in 6G communications and beyond, in which the goal shifts from accurate information transfer at the bit level to the conservation of meaningful information. The limitations in the current state-of-the-art semantic communication frameworks can be summarized as follows: (i) AWGN channel models ignoring fading, phase noise, and hardware imperfections; (ii) reliance on text representations in the form of class labels instead of rich semantics of the natural language; and (iii) fixed error handling techniques without adaptability to channel changes. This paper proposes an UAMM-SC architecture that enables end-to-end transmission of multimodal data over realistic fading channels using adaptive latent correction. Specifically, our contributions include: (1) a cross-attention based encoder to fuse multimodal semantics in a common latent space, with pretrained ViT and DistilBERT as backbone models; (2) a hardware-aware channel model, including Rayleigh fading, Doppler frequency shift, phase noise, and IQ imbalance, simulated using native PyTorch functions;(3) a semantic uncertainty quantification layer based on Monte Carlo Dropout, which triggers gradient-based latent correction adaptively if the predictive entropy is higher than an SNR-adaptive threshold; and (4) an experimental evaluation of the effectiveness of UAMM-SC on CIFAR-10, where we show semantic preservation across SNR ranges (0–20 dB) with a remarkable 99.7% bandwidth reduction compared with JPEG + LDPC transmission.

Zitieren

APA

Shah, M. A., Tiemei, Y., Bashir, K., & Soomro, M. S. (2026). Uncertainty-Aware Multi-Modal Semantic Communication Over 6G Fading Channels with Adaptive Latent Correction. International Journal of Innovative Science and Research Technology, 1254. https://doi.org/10.38124/ijisrt/26jun260

Vancouver

Shah MA, Tiemei Y, Bashir K, Soomro MS. Uncertainty-Aware Multi-Modal Semantic Communication Over 6G Fading Channels with Adaptive Latent Correction. International Journal of Innovative Science and Research Technology. 2026:1254. doi:10.38124/ijisrt/26jun260

BibTeX
@article{Shah2026,
  author = {Shah, Muhammad Afzal and Tiemei, Yang and Bashir, Kashif and Soomro, Muhammad Suleman},
  title = {Uncertainty-Aware Multi-Modal Semantic Communication Over 6G Fading Channels with Adaptive Latent Correction},
  journal = {International Journal of Innovative Science and Research Technology},
  year = {2026},
  pages = {1254},
  doi = {10.38124/ijisrt/26jun260}
}
Kostenlos starten →