Large AI Models-Driven Immersive Communication
IEEE Communications Magazine · 2026-05 · 64 (5) · 132-138
DOI: 10.1109/mcom.001.2500419 ↗Kurz gefasst
Fachartikel in der Zeitschrift IEEE Communications Magazine, 2026-05. Verfasst von Yang Ma und 4 weiteren Autor:innen. Die Arbeit nennt 15 Literaturangaben. Gefördert von: National Natural Science Foundation of China, National Key R&D Program of China.
Abstract (Originalsprache)
The sixth-generation (6G) mobile networks are poised to bridge the physical and virtual worlds, transforming how people work, entertain, and interact. However, delivering high-quality immersive media over 6G entails stringent demands on transmission rate, latency, and reliability, posing significant technical challenges. Recent advancements in large AI models (LAMs) have demonstrated remarkable capabilities in multimodal understanding, large-scale knowledge modeling, and powerful content generation. These capabilities enable LAMs to process and integrate modalities such as text, images, audio, and video, and to generate semantically rich content from compact representations, creating new opportunities for immersive communication. Leveraging these capabilities, we propose a LAMs-driven immersive communication (LAMs-IC) framework that extends the principles of semantic communication for immersive media, ensuring efficient communication with minimal channel overhead. LAMs-IC adopts a hierarchical architecture consisting of cloud, edge, and terminal layers, which collaboratively support semantic abstraction, compression, and multimodal generation. Within this architecture, terminal data is transmitted to the edge layer, where cloud-pretrained LAMs utilize their semantic understanding and generative capabilities to compress information and generate task-specific multimodal content. Consequently, the system facilitates efficient data transmission and high-quality reconstruction, even under bandwidth-constrained conditions, thereby supporting immersive communication. Experimental results demonstrate the superior performance of the LAMs-IC framework in terms of reconstruction quality and symbol efficiency. Finally, we outline key challenges that must be addressed to realize the full potential of LAMs-IC in future 6G immersive media applications.
Förderung
- National Natural Science Foundation of China
- National Key R&D Program of China
Zitieren
APA
Ma, Y., Cheng, W., Wang, J., Zhang, W., & Zhang, H. (2026). Large AI Models-Driven Immersive Communication. IEEE Communications Magazine, 64(5), 132-138. https://doi.org/10.1109/mcom.001.2500419
Vancouver
Ma Y, Cheng W, Wang J, Zhang W, Zhang H. Large AI Models-Driven Immersive Communication. IEEE Communications Magazine. 2026;64(5):132-138. doi:10.1109/mcom.001.2500419
BibTeX
@article{Ma2026,
author = {Ma, Yang and Cheng, Wenchi and Wang, Jingqing and Zhang, Wei and Zhang, Hailin},
title = {Large AI Models-Driven Immersive Communication},
journal = {IEEE Communications Magazine},
year = {2026},
volume = {64},
number = {5},
pages = {132--138},
doi = {10.1109/mcom.001.2500419}
}
EUVIMED