摘要
通过使用聚类分析的方法来进行P2P的流量识别.首先给出P2P流量的特征,接着定义聚类特征树,然后通过扫描从网络中截得的数据得到相应的初始聚类树,最后计算初始聚类的贝叶斯信息准则值得到最终聚类结果.该方法能有效利用存储空间,避免了存储所有数据对象.同时还能够根据数据特征自动得到聚类数目,减少人为因素的影响,与K均值算法相比较优.
Currently P2P traffic identification problem has been a focus and it is a prerequisite for effective management of P2P traffic in order to manage the network better.Traditional identification methods are no longer effective and this paper uses cluster analysis approach to identify P2P traffic.Given the characteristics of P2P traffic,the clustering feature tree is defined,by calculating the data which is scanned from the network,we obtain the corresponding initial tree.The calculation of the BIC of initial clusters is desirable to the final clustering results.This method with the use of storages space is effective,avoiding the storage of the data objects,at the same time it is able to automatically cluster the data by features,and it reduces the influence of human factors,being optimum compared with the K-Means algorithm.
出处
《长沙理工大学学报（自然科学版）》
CAS
2010年第3期58-62,共5页
Journal of Changsha University of Science and Technology:Natural Science
基金
湖南省自然科学基金资助项目(09JJ6094)
