Malicious traffic detection for cloud-edge-end networks: A deep learning approach

Abstract

Malicious traffic has a great impact on network security. This paper studies a malicious traffic detection method based on deep learning. Aiming at the small sample data problem of malicious traffic, it is proposed to enrich the diversity of data by generating an adversarial network so that it can realize the strengthening and expansion of data characteristics and improve the accuracy of data analysis results. The FlowGAN model is used to detect malicious network data and multiple convolutional encoders. Deconvolutional encoders are used to identify malicious traffic. Through the identification and model comparison of different data sets, the AUC values of the model for the three data sets reached 0.9814, 0.7541 and 0.8556, respectively. The model is applied to online system detection. The experimental results show that the average detection accuracy of the method proposed in this paper for malicious traffic reaches 93.89%, which has a good application value.