EUVIMEDEUVIMED entdecken
Crossref

Cross-Domain Semantic Image Communication Empowered by LEO Embodied Intelligence

Yunze Zhang, Chaowei Wang, Wenyuan Wang, Zhi Zhang, Wenjun Xu, Fan Jiang, Lexi Xu

IEEE Transactions on Cognitive Communications and Networking · 2026 · 12 · 8402-8414

DOI: 10.1109/tccn.2026.3695848 ↗

Kurz gefasst

Fachartikel in der Zeitschrift IEEE Transactions on Cognitive Communications and Networking, 2026. Verfasst von Yunze Zhang und 6 weiteren Autor:innen. Die Arbeit nennt 46 Literaturangaben. Gefördert von: National Natural Science Foundation of China.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

To enhance the perception continuity for space-based embodied edge agents in harsh channel environments and the computation-energy balance of advanced generative AI (GenAI) architectures under strict SWaP (Size, Weight, and Power) constraints, this paper proposes a semantic communication framework based on a hybrid Mamba-Transformer architecture. By incorporating Mamba modules characterized by linear computational complexity, the proposed framework effectively captures long-range spatial dependencies inherent in remote sensing images. Furthermore, an implicit minimax optimization strategy is integrated to enhance the intrinsic robustness of the feature space against channel noise, thereby facilitating graceful performance degradation over satellite-to-ground links. Simulation results on diverse datasets demonstrate that the proposed scheme significantly outperforms traditional BPG+LDPC schemes and state-of-the-art deep learning methods in terms of PSNR, MS-SSIM, and LPIPS metrics. Notably, the formulated model exhibits superior cross-domain generalization, enabling the precise reconstruction of critical topological semantics.

Förderung

  • National Natural Science Foundation of China

Zitieren

APA

Zhang, Y., Wang, C., Wang, W., Zhang, Z., Xu, W., Jiang, F., & Xu, L. (2026). Cross-Domain Semantic Image Communication Empowered by LEO Embodied Intelligence. IEEE Transactions on Cognitive Communications and Networking, 12, 8402-8414. https://doi.org/10.1109/tccn.2026.3695848

Vancouver

Zhang Y, Wang C, Wang W, Zhang Z, Xu W, Jiang F, et al. Cross-Domain Semantic Image Communication Empowered by LEO Embodied Intelligence. IEEE Transactions on Cognitive Communications and Networking. 2026;12:8402-8414. doi:10.1109/tccn.2026.3695848

BibTeX
@article{Zhang2026,
  author = {Zhang, Yunze and Wang, Chaowei and Wang, Wenyuan and Zhang, Zhi and Xu, Wenjun and Jiang, Fan and Xu, Lexi},
  title = {Cross-Domain Semantic Image Communication Empowered by LEO Embodied Intelligence},
  journal = {IEEE Transactions on Cognitive Communications and Networking},
  year = {2026},
  volume = {12},
  pages = {8402--8414},
  doi = {10.1109/tccn.2026.3695848}
}
Kostenlos starten →