Optimizations of Flow-based Behavior Recognition System within Hi-speed Network

Since the endless new networking applications and security issues, it's urgent for governments, corporations and ISPs to identify, classify and control the network flows. Comparing with DPI technologies, flow-based behavior recognition has advantages of continuity and extensibility, but has the problems of scalability, efficiency and coordination which greatly restrict its implementation in high-speed core network. Base on related research and experiments, this paper presents optimizations to improve the performance of time and space of flow-based behavior recognition.