2026, Vol. 13, No. 1. - go to content...

Permanent address of this page - https://resources.today/en/05inor126.html

Метаданные этой статьи так же доступны на русском языке

DOI: 10.15862/05INOR126 (https://doi.org/10.15862/05INOR126)

Full article in PDF format (file size: 634.7 KB)

For citation:

Rodygin I.R., Shulzhenko S.N. Applying systems analysis to traffic modeling in fifth-generation mobile networks with integrated machine learning for predictive analytics of non-stationary loads. Russian journal of resources, conservation and recycling. 2026; 13(1). Available at: https://resources.today/PDF/05INOR126.pdf (in Russian). DOI: 10.15862/05INOR126

Applying systems analysis to traffic modeling in fifth-generation mobile networks with integrated machine learning for predictive analytics of non-stationary loads

Rodygin Ivan Romanovich
Moscow State University of Geodesy and Cartography, Korolev, Russia
A.A. Leonov Technological University
E-mail: ivanrrodygin@yandex.ru

Shulzhenko Sergey Nikolaevich
Moscow State University of Geodesy and Cartography, Korolev, Russia
A.A. Leonov Technological University
E-mail: shulzhenko.sn@ut-mo.ru

Abstract. The paper presents an approach to modeling network traffic in fifth-generation mobile networks based on systems analysis and a hybrid forecasting loop. The relevance of the study is driven by non-stationary loads caused by mass Internet of Things connectivity, latency-sensitive services, and the need for proactive resource management. The goal is to improve short-term traffic prediction accuracy and support better allocation decisions under changing arrival rates. The proposed model combines elements of classical queueing theory with machine learning techniques that capture both temporal dynamics and spatial dependencies between base stations. The methodology is evaluated through simulation scenarios with bursty peaks and gradual trends, and the results are compared with baseline statistical and neural approaches. The study shows that integrating a systems perspective with observed data increases robustness of forecasts and reduces the risk of misallocation during peak periods. The paper concludes that the approach is suitable for capacity planning, quality-of-service monitoring, and adaptive management of network slices. Future work includes expanding the feature set and validating the method on real operator data, as well as improving interpretability for engineering decision-making.

Keywords: systems analysis; fifth-generation mobile networks; traffic modeling; non-stationary traffic; queueing theory; graph neural networks; long short-term memory networks; predictive analytics; latency optimization; network resource management

Download article in PDF format