特征选择和支持向量机的P2P网络流量识别模型  

P2P Network Traffic Identification Model Based on Feature Selection and Support Vector Machine

作　　者：刘凯 LIU Kai(Assets Management Corporation,Ltd.Nanjing University of Science and Technology,Nanjing 210094 China)

机构地区：[1]南京理工大学资产经营有限公司,江苏南京210094

出　　处：《自动化技术与应用》2024年第12期98-102,共5页Techniques of Automation and Applications

摘　　要：为了更加有效辨别P2P网络流量,设计特征选择与支持向量机的P2P流量识别模型。将分类错误率和选择特征数最小作为特征选择的两个目标,通过人工蜂群搜索算法去除冗余及无价值特征,精准选取P2P网络流量特征;融合小波分析与支持向量机算法,在依据框架理论并符合Mercer条件下,将Mexican hat小波函数引入支持向量机的核函数,优化支持向量机结构,得到基于小波核函数的支持向量机,以P2P网络流量特征为输入,实现P2P网络流量识别。实验证明:该模型可有效去除P2P流量中多余及无用特征,精准识别P2P网络流量,实用性较强。A P2P traffic identification model based on feature selection and Support Vector Machine is proposed to effectively identify P2P network traffic.Taking the minimum classification error rate and the number of selected features as the two objectives of feature selection,the redundant and worthless features are removed by the artificial bee colony search algorithm,and the P2P network traffic characteristics are accurately selected.Combining wavelet analysis and support vector machine algorithm,according to the framework theory and Mercer conditions,the Mexican hat wavelet function is introduced into the kernel function of support vector machine,and the structure of support vector machine is optimized.The Support Vector Machine based on wavelet kernel function is obtained.Taking the characteristics of P2P network traffic as input,P2P network traffic identification is realized.Experiments show that this model can effectively remove redundant and useless features in P2P traffic,accurately identify P2P network traffic,and has strong practicability.

关 键 词：P2P网络 人工蜂群算法 小波核函数 支持向量机 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构]