EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

Online Reinforcement Policies for Weather-Triggered Vehicular Link Migration Decisions

George Yim

Journal of Innovative Research and Technology · 30.09.2026 · 3 (2) · 203-216

DOI: 10.66857/jirt.v3i2.66 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Journal of Innovative Research and Technology, online erschienen am 30.09.2026. Verfasst von George Yim. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

The rapid evolution of connected and autonomous vehicles relies heavily on robust, high-capacity communication links to support safety-critical applications and high-bandwidth infotainment systems. While millimeter-wave and terahertz frequency bands offer unprecedented data rates, they suffer from severe susceptibility to dynamic meteorological conditions, particularly precipitation, fog, and atmospheric absorption. Traditional rule-based vertical handover mechanisms often fail to adapt to the highly dynamic and unpredictable nature of sudden weather variations, leading to suboptimal link performance, frequent disconnections, and unacceptable latency spikes. This paper proposes a novel framework utilizing online reinforcement learning to optimize weather-triggered vehicular link migration decisions in real-time. By formulating the communication environment as a Markov Decision Process, the proposed system enables a learning agent to continuously observe local meteorological data, current link quality indicators, and vehicular mobility states to make proactive link migration decisions between high-frequency and highly reliable sub-6 gigahertz backup bands. Unlike offline learning models that struggle with unseen weather anomalies, the online reinforcement learning policy continuously updates its action-value estimations, balancing the trade-off between maximizing instantaneous throughput and minimizing the signaling overhead associated with frequent handovers. Extensive experimental evaluations using synthesized meteorological traces and mobility datasets demonstrate that the proposed online policy significantly reduces packet drop rates and maintains higher average throughput during severe weather events compared to baseline static threshold methods and offline learning approaches.

Zitieren

APA

Yim, G. (2026). Online Reinforcement Policies for Weather-Triggered Vehicular Link Migration Decisions. Journal of Innovative Research and Technology, 3(2), 203-216. https://doi.org/10.66857/jirt.v3i2.66

Vancouver

Yim G. Online Reinforcement Policies for Weather-Triggered Vehicular Link Migration Decisions. Journal of Innovative Research and Technology. 2026;3(2):203-216. doi:10.66857/jirt.v3i2.66

BibTeX
@article{Yim2026,
  author = {Yim, George},
  title = {Online Reinforcement Policies for Weather-Triggered Vehicular Link Migration Decisions},
  journal = {Journal of Innovative Research and Technology},
  year = {2026},
  volume = {3},
  number = {2},
  pages = {203--216},
  doi = {10.66857/jirt.v3i2.66}
}
Kostenlos starten →