SAE-CNN: A Spatial Attention Augmented Convolutional Neural Network for Service Type Identification in 5G Heterogeneous Networks

As the rapid expansion of 5G heterogeneous networks brings increasingly complex traffic patterns, effective traffic identification becomes crucial to maintaining optimal network management and enhancing user experience. To address this challenge, this paper proposes a spatial attention-augmented convolutional neural network (SAE-CNN) for traffic identification in 5G heterogeneous networks. By incorporating the spatial attention mechanism, the SAE-CNN accurately processes network traffic data and improves the accuracy and efficiency of classification. In the evaluation of a simulated 5G environment and the ISCXTor2016 dataset, the model shows 98.72% accuracy, which is a 1.17% improvement over existing methods. In addition, we deploy the proposed model in a 5G environment for online dynamic testing, and the experimental results further confirm its effectiveness in real-world 5G network scenarios.