适用于资源受限设备的移动应用类别实时识别方法  

THE METHOD OF MOBILE APPLICATION REAL-TIME IDENTIFICATION FOR RESOURCE-CONSTRAINED DEVICES

作　　者：陈旖 张美璟 许发见[1] Chen Yi;Zhang Meijing;Xu Fajian(Department of Computer and Information Security Management,Fujian Police College,Fuzhou 350007,Fujian,China)

机构地区：[1]福建警察学院计算机与信息安全管理系,福建福州350007

出　　处：《计算机应用与软件》2021年第7期80-86,92,共8页Computer Applications and Software

基　　金：教育部人文社科研究青年项目(17YJC630213);福建省自然科学基金面上项目(2017J01514);福建警察学院厅级课题(JW201907)。

摘　　要：针对基于流量分析的移动应用类别识别方法存在计算量大、难以实时识别的问题,提出一种移动应用实时识别方法。根据应用访问域名的特征,将报文进行转换和降维来生成样本向量,并使用支持向量机进行分类。在微型无线网关上对其测试,在对一组目标应用进行识别时,该方法的识别准确率约为94.4%,CPU使用率峰值约1.8%,内存消耗约1052 KB,吞吐量略微降低。实验表明,该方法能满足资源受限的网络设备进行移动应用类别实时识别的需求。The problem of the mobile application identification based on flow analysis is intensive computation and difficulty in real-time identification.In view of this,a real-time identification method for mobile applications is put forward.The method converted and reduced the dimension of the message samples to get the corresponding vectors according to the feature of applications access domain name.Then it used the support vector machine for classification.The method was tested in a micro wireless gateway.The test results show that the recognition accuracy of the method is about 94.4%,the CPU usage peak is about 1.8%,the memory consumption is about 1052 KB,and the throughput is slightly reduced when identifying a set of target applications.The experimental results show that the identification method can meet the needs of real-time identification of mobile applications for resource-constrained network devices.

关 键 词：手机应用类别识别 数据流挖掘 Jaccard包相似度 支持向量机 布隆过滤器 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]