Research on traffic representation in network anomaly detection

Abstract

Aiming to address the problem of information loss in traffic representation for network anomaly detection, the impact of feature information dimension of different traffic representation on anomaly detection performance was analyzed from the perspective of data collection granularity. Firstly, the integrated collaboration between traffic representation granularity and the coupling among traffic representation, feature learning, and detection in malicious anomaly detection was introduced. Subsequently, the evolution of traffic representation in network anomaly detection was systematically reviewed, providing a comprehensive analysis of its forms, feature learning, and application in anomaly detection both globally and domestically. Finally, the future research directions revolving around the collaborative development trend of traffic representation in network anomaly detection were outlined.

Keywords

• anomaly detection
• network traffic
• traffic representation form
• feature type
• multimodal traffic representation