基于最优ABC-SVM算法的P2P流量识别  

Identification of P2P traffic based on optimal ABC-SVM

作　　者：王春枝[1] 杜远丽 叶志伟[1] 

机构地区：[1]湖北工业大学计算机学院,武汉430068

出　　处：《计算机应用研究》2018年第2期582-585,共4页Application Research of Computers

基　　金：国家自然科学基金资助项目(61170135)

摘　　要：目前对等网络(peer-to-peer,P2P)流量的识别是网络管理研究的热门话题。基于支持向量机(support vector machine,SVM)的P2P流量识别方法是常用的P2P流量识别方法之一。然而SVM的性能主要受参数和其使用特征的影响,传统的方法是将SVM的参数优化和特征选择问题分开处理,但是难以获得整体性能最优的SVM分类器。针对以上问题进行了研究,提出了一种基于最优人工蜂群算法与支持向量机相结合的P2P流量识别方法。利用人工蜂群算法,将SVM的参数和特征选择问题视为最优化问题同步处理,可以获得整体性能最优的参数和特征子集。在真实的P2P数据上的实验结果表明,提出的方法具有很好的自适应性和分类精度,能够同时获取特征子集和SVM参数的最优解,提高SVM分类器的整体性能。Currently peer-to-peer （P2P） network traffic identification is a hot topic in network management. Identification of P2P traffic based on support vector machine （SVM） is a commonly used P2P traffic identification method. However, the per- formance of SVM is mainly affected by the parameters and features used, the traditional method is to optimize the parameters and features of SVM separately. Hence, it is difficult to obtain the optimal SVM classifier on the whole. This paper proposed a P2P traffic identification approach based on artificial bee colony algorithm and the optimal SVM. Tuning parameters of SVM and feature selection was regarded as the optimization problem, which was handled with artificial bee colony algorithm synchro- nously. As a result, it obtained the optimal parameters and feature subset of SVM. The results show that the proposed method has good adaptability and classification accuracy on the real P2P data; it can simultaneously obtain the optimal feature subset and parameters of SVM and improve the overall performance.

关 键 词：人工蜂群算法 支持向量机 特征选择 参数优化 P2P流量识别 

分 类 号：TP393.07[自动化与计算机技术—计算机应用技术]