AI-based Traffic Forecasting in 5G network

Forecasting of the telecommunication traffic is the foundation for enabling intelligent management features as cellular technologies evolve toward fifth-generation (5G) technology. In this work, a deep-learning based analysis of a traffic dataset was conducted. For this purpose, several neural network-based models are utilized. The paper explores the forecasting performance of the fully connected sequential network (FCSN). Specifically, one-dimensional convolutional neural network (1D-CNN), single shot learning LSTM (SS-LSTM), and autoregressive LSTM (AR-LSTM) models have been evaluated. In addition, the baseline model was developed to assess the performance of the aforementioned models. The results reveal that FCSN and 1D-CNN have comparable performance. However, 1D-CNN is a smaller network with less number of parameters. One of the other benefits of the proposed 1D-CNN is having less complexity and faster execution time for predicting the next 24-hour traffic.