EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

A Cross-Layer Deep Reinforcement Learning-Based Framework for STAR-RIS-Assisted MISO Video Communications

Yu-Chun Huang, Shu-Ming Tseng

IEEE Open Journal of the Communications Society · 2026 · 7 · 6508-6526

DOI: 10.1109/ojcoms.2026.3699518 ↗

Kurz gefasst

Fachartikel in der Zeitschrift IEEE Open Journal of the Communications Society, 2026. Verfasst von Yu-Chun Huang und 1 weiteren Autor:innen. Die Arbeit nennt 55 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: National Science and Technology Council, Taiwan.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

The rapid proliferation of multimedia services has placed stringent requirements on wireless networks to simultaneously support high data rates and consistent perceptual video quality. To address coverage limitations and unfavorable propagation conditions in beyond-5G and emerging 6G systems, Simultaneously Transmitting and Reflecting Reconfigurable Intelligent Surfaces (STAR-RIS) have been introduced as a flexible solution capable of reshaping wireless channels over the entire spatial domain. Despite their potential, most existing STAR-RIS designs remain confined to physical-layer performance objectives and adopt complex deep reinforcement learning (DRL) algorithms, leaving the interaction between wireless resource allocation and application-layer video characteristics largely unexplored. In this work, we develop a cross-layer optimization framework for STAR-RIS-assisted multi-user MISO video communication that explicitly targets end-to-end video quality. A Proximal Policy Optimization (PPO)-based learning strategy is employed to jointly optimize base station beamforming, STAR-RIS transmission and reflection coefficients, and video encoding control, while maintaining a lightweight and stable training structure. By embedding video rate–distortion parameters into the learning state and directly maximizing Peak Signal-to-Noise Ratio (PSNR), the proposed approach enables content-aware STAR-RIS configuration with improved robustness. In addition, the proposed framework is extended to a STAR-RIS-assisted mobile edge computing (MEC) environment using a two-stage learning architecture that decouples wireless configuration from computation offloading decisions. Simulation results demonstrate that the proposed cross-layer scheme achieves an average PSNR gain of approximately 3-4 dB in downlink video transmission, and an additional improvement of 1-3 dB in MEC offloading scenarios, confirming its effectiveness for multimedia-oriented wireless systems.

Förderung

  • National Science and Technology Council, Taiwan

Zitieren

APA

Huang, Y., & Tseng, S. (2026). A Cross-Layer Deep Reinforcement Learning-Based Framework for STAR-RIS-Assisted MISO Video Communications. IEEE Open Journal of the Communications Society, 7, 6508-6526. https://doi.org/10.1109/ojcoms.2026.3699518

Vancouver

Huang Y, Tseng S. A Cross-Layer Deep Reinforcement Learning-Based Framework for STAR-RIS-Assisted MISO Video Communications. IEEE Open Journal of the Communications Society. 2026;7:6508-6526. doi:10.1109/ojcoms.2026.3699518

BibTeX
@article{Huang2026,
  author = {Huang, Yu-Chun and Tseng, Shu-Ming},
  title = {A Cross-Layer Deep Reinforcement Learning-Based Framework for STAR-RIS-Assisted MISO Video Communications},
  journal = {IEEE Open Journal of the Communications Society},
  year = {2026},
  volume = {7},
  pages = {6508--6526},
  doi = {10.1109/ojcoms.2026.3699518}
}
Kostenlos starten →