Deep Learning for IoT Traffic Prediction based on Edge Computing

Ali R. Abdellah, Artem Volkov, Ammar Muthanna, Andrey Koucheryavy
The 5G network is the latest wireless mobile communication technology. Nowadays, the emerging of many network applications has led to a massive amount of network traffic. Many researchers have devoted their studies to the accurate prediction of network traffic applications. Network management requires technology for the prediction of network traffic without network operator intervention. In practice, 5G uses the Internet of Things (IoT) for working in high-traffic networks with multiple sensors to send their packets to a destination simultaneously, which is a feature of IoT applications. Therefore, 5G offers wide bandwidth, low delay, and extremely high data throughput. Predicting network traffic is more important for IoT networks to provide reliable communication. The efficient 5G network cannot be complete without including artificial intelligence (AI) procedures. Machine learning (ML) has been successfully applied to traffic prediction. In this paper, we implement the prediction of IoT traffic in time series using deep learning. the prediction accuracy has been evaluated using the (RMSE) as a merit function. and mean absolute percentage of error (MAPE).