基于3G核心网数据的用户网络行为识别技术研究

发布时间：2019-06-17 21:40

【摘要】：3G网络的普及加速了人们迈向互联网时代的步伐,通信运营商拥有庞大的3G用户数量,越来越多的研究人员加入到3G网络数据分析与研究的行列中。但是目前对于3G网络数据的研究主要是基于运维数据的用户移动行为的分析与研究,很少涉及到用户的网络行为。在当今互联网的大时代下研究用户的网络行为有着十分重要的意义。通过研究移动用户的网络行为,运营商除了能够持续跟踪记录用户的网络行为、探索用户深层次的业务需求,还可以将研究的结果用来指导制定运营商的营销策略,真正的实现面向用户的精准营销。在对用户网络行为的研究中,本文首先利用采集的3G核心网数据实现了移动应用的识别,接着设计实现了用户网络行为的识别与预测系统,较为全面地分析与挖掘了用户的网络行为,本文的研究工作主要分为以下几部分:1.本文首先介绍了用户网络行为研究的目的以及国内外关于此课题取得的一些研究成果;接着介绍了3G核心网数据的采集与预处理,确定了数据的采集接口并给出了源数据预处理的实现过程,得到了可供分析使用的数据集。2.实现了移动应用的识别与用户网络行为偏好的分析。通过DPI技术,本文分析了目前主流的移动应用并且提取出了每种移动应用的特征值,然后根据这些特征值建立了移动应用特征库,通过采集数据与特征库的匹配识别出用户使用的移动应用;在移动应用的识别基础上,本文提出了基于分散度识别算法的用户网络行为偏好分析系统,对用户的上网偏好进行了分析,实现了对移动用户上网时段与上网内容偏好的识别。3.本文通过决策树算法建立了网络行为的预测模型,实现了用户网络行为的预测;通过对比常用的决策树算法,本文选择C4.5算法建立决策树,同时给出了处理连续型属性的方法,完成了连续型属性的离散化过程;接着通过后剪枝技术实现了对预测模型的剪枝与优化,并给出了详细的实现方法,最后得到了较为可靠的网络行为预测模型。
[Abstract]:The popularization of the 3G network has accelerated the pace of people to the Internet era, and the communication operators have a large number of 3G users, and more and more researchers have joined the 3G network data analysis and research. However, the research on the 3G network data is mainly based on the analysis and research of the user's mobile behavior based on the operation and maintenance data, and the network behavior of the user is seldom involved. It is of great significance to study the network behavior of users in the large era of the Internet. By studying the network behavior of the mobile user, the operator can continuously track and record the network behavior of the user, explore the user's deep-level business demand, and also can use the results of the research to guide the development of the operator's marketing strategy, and the real realization of the user-oriented accurate marketing. In this paper, the identification of the mobile application is realized by using the acquired 3G core network data, and then the identification and prediction system of the user's network behavior is designed, and the network behavior of the user is analyzed in a more comprehensive way. The research work in this paper is divided into the following parts:1. This paper first introduces the purpose of user network behavior research and some research results on this subject at home and abroad, then introduces the acquisition and pre-processing of 3G core network data, determines the data collection interface and gives the realization process of source data pre-processing. A set of data available for analysis is obtained. And the analysis of the identification of the mobile application and the user's network behavior preference is realized. Through the DPI technology, this paper analyzes the current mobile application and extracts the characteristic value of each mobile application, then sets up the mobile application feature library according to the characteristic values, and recognizes the mobile application to be used by the user through the matching of the collected data and the feature library; On the basis of the identification of mobile applications, a user's network behavior preference analysis system based on the dispersity identification algorithm is proposed, and the user's online preference is analyzed. In this paper, the prediction model of network behavior is established by the decision tree algorithm, and the prediction of user's network behavior is realized. By comparing the common decision tree algorithm, the C4.5 algorithm is chosen to set up the decision tree, and the method of processing the continuous property is given. In this paper, the discrete process of the continuous attribute is completed, and the pruning and optimization of the prediction model are realized by the post-pruning technique, and the detailed implementation method is given, and a reliable network behavior prediction model is finally obtained.
【学位授予单位】：电子科技大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TN929.53

【相似文献】