Network Abnormal Traffic Detection Framework Based on Deep Reinforcement Learning
Shi Dong, Yuanjun Xia, Tao Wang
Abstract
Because wireless communication network attacks continue to evolve, cyber security is becoming more and more important, requiring network security solutions to be constantly updated. In addition, wireless communication network traffic in high-speed networks is massive, high-dimensional, dynamic, and so on, making attacks difficult to detect in real-time. Unknown attacks are also difficult to identify. This article proposes a framework for abnormal traffic detection based on deep reinforcement learning, mainly consisting of four stages: data collection, dataset pre-processing, feature selection, and model detection. Experimental results show that the abnormal traffic detection framework has improved detection performance.