Application of Machine Learning in Predicting Bandwidth Requirements on The 5G Networks

This research explores the application of Machine Learning algorithms, specifically Long Short-Term Memory (LSTM), in predicting bandwidth requirements in 5G networks. In this study, three Machine Learning models are evaluated, namely LSTM, Random Forest, and Support Vector Machine (SVM), to measure their performance in estimating bandwidth requirements based on network traffic data. The dataset used includes 100,000 data points collected over one week. The results showed that the LSTM model produced the most accurate prediction, with a Mean Absolute Error (MAE) of 0.034 and a Root Mean Square Error (RMSE) of 0.056. The Random Forest model showed lower performance, with MAE of 0.062 and RMSE of 0.098, while the SVM gave the worst results with MAE of 0.073 and RMSE of 0.112. The application of LSTM in 5G networks enables more accurate bandwidth requirement prediction and can be integrated into smart grid systems to improve bandwidth allocation efficiency and quality of service (QoS). This research also suggests the use of larger datasets as well as the application of hybrid models in the future to improve prediction accuracy.