基于深度检测的用户行为分析以及 Web 推送的设计与实现

摘要

With the rapid development of the Internet and the advanced technology such as data mining, web server push technology has been constantly brought forth the fresh and become variety. But in this paper a new idea of web push is pro-vided. The new method takes the advantages of MIPS network dedicated processors and get rid of the limit of server. It uses the Deep Packet Inspection at the gateway of whole local network to analysis user's behavior and offers real-time custo-mized web push to the user directly. First, the paper introduces the basic framework and advantages of the MIPS network dedicated processors in user's behavior analysis.Then the unique operation mechanism of web push is shown secondly. And the basic framework of the analysis and web push system is demonstrated next. The details of the system is displayed at last. This system successfully implements the user's behavior analysis and customized web push function without the limit of server and works with a high performance.%随着互联网的高速发展以及对数据挖掘等先进技术的应用，web 端服务器推送技术不断推陈出新，变得种类繁多。而本文从另一个角度出发，为 web 推送的方法提供了一套新的思路。利用 MIPS 架构多核专用网络处理器，通过对整个网络流量的数据深度检测，结合用户行为分析，摆脱服务器限制，直接向用户提供灵活的 web 推送服务。本文首先介绍了 MIPS 架构专用网络处理器在用户行为分析方面的运行机制以及优点。其次介绍了系统在 web 推送方面与传统服务端推送的不同以及详细机制。随后展示了整个系统的整体框架以及基本任务分割。最后对系统的具体设计和实现细节进行了详细的介绍。本文提出的用户行为分析以及 web 推送系统，充分利用了专用网络处理器的硬件性能，避开了服务器限制，在保障网络通信高效畅通的前提下，实现了实时灵活的 web 推送机制。