Mobile-based Network Traffic Analysis Technology And Management System Implementation

With the rapid popularization of mobile devices and the rapid development of mobile communication technology,mobile Internet traffic has already accounted for the vast majority of network traffic.Mobile phones have changed from communication equipment to people's living equipment,entertainment equipment and payment equipment.Currently,mobile network traffic lacks management,and there are still malicious traffic and automatically generated background traffic on mobile devices.For network administrators,ensuring user information security and improving user experience are the primary issues.The particularity and complexity of mobile network traffic have brought great challenges to traditional traffic collection methods and analysis techniques,and traditional methods seem to be stretched.First,mobile devices are not suitable for collecting and labeling large amounts of traffic data.Secondly,with the emergence of dynamic port technology and traffic encryption technology,the classic port and packet load content analysis methods are not effective;the machine learning classification method based on statistical features requires experts to design the extracted features,and limited human resources cannot Deal with the explosive growth of traffic data.Finally,the existing mobile traffic management systems are only designed and developed for researchers,and their functions are not perfect and are not suitable for ordinary users.To this end,this paper carries out research on mobile traffic analysis technology based on deep learning,designs and implements a mobile network traffic management system.The main tasks are as follows:(1)In order to effectively collect and analyze mobile traffic,this article designs a collection method based on VPN and switch port analyzer,combined with monitoring of Android system files,to obtain traffic data and label information.On this basis,the characteristic analysis and comparison of the load length and time interval of the traffic data are carried out.Through the similarity analysis of the picture structure,it is found that the load length characteristic is more obvious.(2)In order to compare the ability of different deep learning models to extract different traffic characteristics,this paper conducts deep learning comparative experiments based on the above two kinds of traffic data.Through the analysis of experimental data,it is found that the depth characteristics based on the length of the load can better reflect the characteristics of mobile traffic.Comparing different models using load length characteristics,it is concluded that the performance of the VGG16 network model is better.(3)Combining the contents of work(1)and(2),this article builds a mobile traffic monitoring and management App system.The system includes two modules: mobile and server.The mobile terminal module has the functions of traffic monitoring and statistics,connection information display and network authority.The server has traffic forwarding,collection and analysis functions.The system proposed in this paper has more complete functions than the existing traffic management system,and is applicable to a wider audience and has stronger practicability.This article's research on mobile traffic collection,analysis and management is of significance to network administrators,researchers and ordinary users,and it provides an effective method for network security and management.