EUVIMEDEUVIMED entdecken
Crossref

Deep reinforcement learning-based performance optimization for sparse irregular STAR-RIS-assisted multi-user MIMO systems

Jianbin Xue, Zhenqin Wang, Xiao Li, Hefei Wu

Physica Scripta · 26.06.2026 · 101 (25) · 256006

DOI: 10.1088/1402-4896/ae776f ↗

Kurz gefasst

Fachartikel in der Zeitschrift Physica Scripta, online erschienen am 24.06.2026. Verfasst von Jianbin Xue und 3 weiteren Autor:innen. Die Arbeit nennt 23 Literaturangaben. Gefördert von: Graduate Innovation Star, Gansu Provincial Department of Education.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

To reduce the hardware cost and control power of conventional fully activated reconfigurable intelligent surfaces (RISs), this paper investigates a sparse irregular simultaneously transmitting and reflecting RIS (STAR-RIS)-assisted downlink multi-user multiple-input multiple-output system. In the considered architecture, STAR-RIS elements are selected from a predefined grid, and only part of the candidate grid points are activated to form an irregular sparse topology. The transmission and reflection coefficients of the STAR-RIS and the beamforming matrices at the base station are jointly optimized under imperfect channel state information (CSI). To handle the resulting high-dimensional continuous optimization problem, a deep deterministic policy gradient-based framework is developed, where the actor network outputs the beamforming and STAR-RIS coefficient variables, and normalization is applied to satisfy the transmit-power and energy-splitting constraints. Numerical simulations show that the proposed irregular sparse STAR-RIS scheme achieves higher sum rate and energy efficiency (EE) than regular full and regular sparse baselines under different numbers of active elements, transmit powers, path loss exponents, Rician factors, and CSI conditions. The results indicate that irregular sparse deployment can improve hardware utilization and EE while maintaining competitive transmission performance.

Förderung

  • Graduate Innovation Star
  • Gansu Provincial Department of Education

Zitieren

APA

Xue, J., Wang, Z., Li, X., & Wu, H. (2026). Deep reinforcement learning-based performance optimization for sparse irregular STAR-RIS-assisted multi-user MIMO systems. Physica Scripta, 101(25), 256006. https://doi.org/10.1088/1402-4896/ae776f

Vancouver

Xue J, Wang Z, Li X, Wu H. Deep reinforcement learning-based performance optimization for sparse irregular STAR-RIS-assisted multi-user MIMO systems. Physica Scripta. 2026;101(25):256006. doi:10.1088/1402-4896/ae776f

BibTeX
@article{Xue2026,
  author = {Xue, Jianbin and Wang, Zhenqin and Li, Xiao and Wu, Hefei},
  title = {Deep reinforcement learning-based performance optimization for sparse irregular STAR-RIS-assisted multi-user MIMO systems},
  journal = {Physica Scripta},
  year = {2026},
  volume = {101},
  number = {25},
  pages = {256006},
  doi = {10.1088/1402-4896/ae776f}
}
Kostenlos starten →