基于聚类分析的高性能包分类技术研究

包分类技术是路由器和防火墙等网络设备的核心技术，也是制约计算机网络发展的关键技术，具有重要的研究意义和应用前景。针对其面临的高性能挑战，如支持线速数据包处理、高可扩展性和OpenFlow网络等，本课题展开基于聚类分析的高性能包分类技术研究。首先，引入数据挖掘中的聚类分析理论，对包分类问题进行严格的形式化建模；其次，利用规则的聚集特征，以规则聚类为切入点，设计高效的包分类结构、算法和机制，重点研究基于空间分布密度聚类的、基于无前缀关系聚类的、基于前缀位聚类的、基于前缀长度聚类的高性能包分类算法；最后，通过“三段迭代式”的实验方案，使研究成果最终走向实际应用。本项目从理论研究出发，展开数据结构和算法的设计，最终走向系统部署的应用和优化，项目的研究成果能为相应网络设备的研发或者其他相关工作的开展提供参考和经验，具有良好学术价值和应用前景。

Packet classification technology is the key technology of network devices like routers and firewalls. It is also the key technology which restricts the development of computer networks. Thus, studying the packet classification technology has vital research significance and application prospects. This project studies the high-performance packet classification technology based on clustering analysis in order to solve high-performance problems such as supporting wire-speed packet processing, high scalability and OpenFlow network. First, we introduce the clustering analysis theory of data mining, and strictly model the problems of packet classification. Second, we design efficient packet classification structures, algorithms and mechanisms by utilizing the aggregation characteristics of rules. This project focuses on the high-performance packet classification algorithms based on the density of space distribution clustering, no prefix relationship clustering, prefix bit clustering and prefix length clustering. Finally, we realize the practical application of research results through the “three-stage iterative” experiment program. This project starts with the theoretical studies, and then designs the data structure and algorithm, and ultimately realizes the applications and optimization of system deployment. The research results of this project provide reference and experience for the development of network devices or other related works. This project has good academic value and application prospects.