Прогнозування завантаженості каналів у бездротових локальних мережах

Abstract

Wireless Local Area Networks (WLAN) play a crucial role in modern communication systems. However, network congestion, signal instability, and bandwidth limitations remain key challenges. Predictive modeling provides an effective solution by analyzing historical data and forecasting network conditions. This paper reviews various forecasting methods, including time series analysis (ARIMA, ETS), machine learning (Decision Trees, Random Forest, XGBoost), and deep learning (RNN, LSTM, CNN). The study highlights the advantages and limitations of these models and explores hybrid approaches for improved accuracy. The findings suggest that predictive models can significantly enhance WLAN performance, optimizing bandwidth allocation and reducing latency.