WF-C4.5:WiFi环境下基于C4.5决策树的手持终端流量识别方法
发布时间：2018-11-06 12:40
【摘要】：目前移动数据流量已占全球IP流量的47%,其中WiFi流量已占整个移动数据流量的90%以上。WiFi环境下移动终端流量的识别对互联网流量管理具有重要意义。传统基于HTTP用户代理(User Agent,UA)的流量识别方法存在识别率不高的问题。分析了WiFi环境下移动终端连接持续时间、数据包大小、有效载荷大小等流量特征,提出一种WiFi环境下基于C4.5决策树的手持终端设备流量识别方法 WF-C4.5,通过计算各属性值的信息增益率构建决策树模型,实现手持终端与非手持终端流量的区分。实验表明,相比UA方法65%的准确率,所提方法的准确率高达95%。
[Abstract]:At present, mobile data traffic has accounted for 47% of global IP traffic, of which WiFi traffic has accounted for more than 90% of the total mobile data traffic. The identification of mobile terminal traffic in WiFi environment is of great significance to Internet traffic management. The traditional traffic identification method based on HTTP user Agent (User Agent,UA) has the problem of low recognition rate. This paper analyzes the flow characteristics of mobile terminal connection duration, packet size and payload in WiFi environment, and proposes a mobile terminal traffic identification method WF-C4.5, based on C4.5 decision tree in WiFi environment. By calculating the information gain rate of each attribute value, a decision tree model is constructed to distinguish the traffic between the handheld terminal and the non-handheld terminal. The experimental results show that the accuracy of the proposed method is as high as 95% compared with the 65% accuracy of the UA method.
【作者单位】： 湖北大学计算机与信息工程学院;湖北省教育信息化工程技术研究中心;
【基金】：赛尔网络下一代互联网技术创新项目(NGII20150101)资助
【分类号】：TN92;TP393.06
，
本文编号：2314298
[Abstract]:At present, mobile data traffic has accounted for 47% of global IP traffic, of which WiFi traffic has accounted for more than 90% of the total mobile data traffic. The identification of mobile terminal traffic in WiFi environment is of great significance to Internet traffic management. The traditional traffic identification method based on HTTP user Agent (User Agent,UA) has the problem of low recognition rate. This paper analyzes the flow characteristics of mobile terminal connection duration, packet size and payload in WiFi environment, and proposes a mobile terminal traffic identification method WF-C4.5, based on C4.5 decision tree in WiFi environment. By calculating the information gain rate of each attribute value, a decision tree model is constructed to distinguish the traffic between the handheld terminal and the non-handheld terminal. The experimental results show that the accuracy of the proposed method is as high as 95% compared with the 65% accuracy of the UA method.
【作者单位】： 湖北大学计算机与信息工程学院;湖北省教育信息化工程技术研究中心;
【基金】：赛尔网络下一代互联网技术创新项目(NGII20150101)资助
【分类号】：TN92;TP393.06
，
本文编号：2314298
本文链接：https://www.wllwen.com/kejilunwen/xinxigongchenglunwen/2314298.html
