Прогнозування та оптимізація інформаційного навантаження в телекомунікаційних мережах

Abstract

The paper considers approaches to forecasting and optimizing information load in next-generation telecommunication networks. It is shown that the increasing volume and heterogeneity of traffic in heterogeneous systems complicate resource management and the provision of the required Quality of Service (QoS) level. To enhance network performance efficiency, the use of hybrid forecasting models combining statistical (ARIMA, GARCH) and intelligent (LSTM, DNN) approaches is proposed. Short-term and long-term forecasting methods are analyzed, and the influence of traffic self-similarity and multifractality on prediction accuracy is evaluated. The modeling results demonstrate that combined models can reduce the mean square forecasting error by 10– 15% compared to traditional statistical approaches. The proposed methods can be applied for intelligent resource management, load balancing, and improving the operational stability of 5G and 6G networks.