EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

Machine Learning Enabled Congestion Prediction and Traffic Steering in 5G Radio Access Networks

Mohammad Akhtar Khan

Journal of Intelligent Decision Making and Information Science · 05.09.2026 · 3 (17s) · 2836-2843

DOI: 10.17762/jidmis.v3.5691 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Journal of Intelligent Decision Making and Information Science, online erschienen am 05.09.2026. Verfasst von Mohammad Akhtar Khan. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Traffic congestion in fifth-generation radio access networks is spatially uneven, time varying and increasingly service dependent. Static utilization thresholds identify congestion only after it has affected throughput, latency and session continuity. This review develops a machine-learning framework that predicts cell-level congestion and converts forecasts into governed traffic-steering decisions. The design combines performance counters, radio conditions, mobility flows, topology, configuration, calendar and event information. It distinguishes demand congestion from radio-quality, interference, transport and hardware limitations so that steering is applied only when a suitable target layer or neighboring cell exists. Statistical baselines, gradient-boosted trees, recurrent models and graph-based approaches are compared according to accuracy, interpretability, inference time and operational risk. A two-stage decision process first estimates congestion probability and severity over several horizons and then selects a bounded steering action subject to capacity, mobility, coverage and quality guardrails. The framework is adapted to Saudi Arabia, where metropolitan growth, pilgrimage seasons, stadium events, transport corridors, industrial locations and giga-projects create exceptional variations in mobile demand. It includes shadow-mode testing, human approval, canary deployment, rollback and post-change causal validation. The study concludes that machine learning can shift RAN management from reactive threshold handling to proactive assurance, but reliable gains depend on clean temporal data, peer-group baselines, calibrated uncertainty and disciplined change governance.

Zitieren

APA

Khan, M. A. (2026). Machine Learning Enabled Congestion Prediction and Traffic Steering in 5G Radio Access Networks. Journal of Intelligent Decision Making and Information Science, 3(17s), 2836-2843. https://doi.org/10.17762/jidmis.v3.5691

Vancouver

Khan MA. Machine Learning Enabled Congestion Prediction and Traffic Steering in 5G Radio Access Networks. Journal of Intelligent Decision Making and Information Science. 2026;3(17s):2836-2843. doi:10.17762/jidmis.v3.5691

BibTeX
@article{Khan2026,
  author = {Khan, Mohammad Akhtar},
  title = {Machine Learning Enabled Congestion Prediction and Traffic Steering in 5G Radio Access Networks},
  journal = {Journal of Intelligent Decision Making and Information Science},
  year = {2026},
  volume = {3},
  number = {17s},
  pages = {2836--2843},
  doi = {10.17762/jidmis.v3.5691}
}
Kostenlos starten →