Multi-UPF Traffic Prediction of 5G Core Network With Multi-Task Learning and Multi-core Parallel Computing

Traffic prediction has become a key aspect of 5G core network management. However, predicting traffic for multiple User Plane Functions (Multi-UPF) has received limited attention. To achieve effective and efficient forecasting for Multi-UPF traffic with machine learning, this study proposes a framework combining multi-task learning and multi-core parallel computing. Firstly, the time-series traffic data of Multi-UPF is concatenated to form the input feature. Then, Random Forest algorithm is used as the backbone of the multi-task learning-based prediction model, which is able to predict the traffic of each UPF simultaneously. Finally, the computation of the multi-task learning model is further accelerated through multi-core parallel processing. Experimental results demonstrate that the proposed multi-task learning approach outperforms single-task learning method in the accuracy of each UPF traffic prediction. Furthermore, the use of multi-core parallel computing enhances the prediction speed significantly.