针对网络服务提供商的一种流量分类和应用架构

数据流量识别技术在应对网络病毒、网络入侵检测和网络管理等方面部有着重要的应用。当前主要的网络流量识别技术主要是DPI（Deep Packet Inspection）和DFI（Deep Flow Inspection），两种技术都各自有优点和缺点。本史提出了一种网络架构，将DPI和DFI技术相结合，并通过数据挖掘技术来辅助发现分类规则，应用FPGA的可编程性和并发性来加速识别和处理过程。

A Traffic Classification and Application Network Framework for ISPs

Network traffic classification provides important benefits for IP network intrusion detecting, management and control. Current popular classification methods are Deep Packet Inspection （DPI） and Deep Flow Inspection （DFI）, but both have disadvantages. This paper introduces a classification framework which combines DPI and DFI. Process rules based on classification are implemented in routers, and rules mining servers are used to find classification rules.