Design And Implementation Of Mobile Internet Behavior Audit System

In recent years, the great progress of information technology makes the mobile Internet based on3G/4G/LTE and WLAN gradually become the main information delivery platform. The daily growth rate of mobile Internet users is more than the traditional Internet users. Different from the traditional Internet, mobile Internet has extremely strict requirements to protect user privacy and user behavior, it has higher security requirements than traditional Internet. Mobile Internet is based on the traditional Internet threats with security threats against mobile properties.And mobile Internet security audit is one of the network security technology to solve this problem emerge as the times require.Different with the traditional Internet security audit is that it mainly carries on the security audit according to the online behavior of mobile terminal, and with high real-time requirements. This paper mainly analyzes the behavior audit technology of mobile Internet security audit.After studying the theoretical basis about the key technologies of behavior audit, this paper combined the semi-supervised machine learning and the selective ensemble learning technology through analysis and comparison, and established an abnormal behavior detection model based on selective semi-supervised learning. The model is successfully applied to the mobile Internet behavior audit system in this article, and implements the security audit about mobile Internet users behavior. Research work of this paper mainly contains the following areas:(1)This paper carried on a detailed investigation about the development of mobile Internet, and did an analysis of some existing security hidden danger of mobile Internet. At the same time, it analyzed several typical security audit products at home and abroad. On the base of depth analysis of mobile Internet security audit products, the importance and urgency of the study of mobile Internet security audit products is discussed.(2)The article discuss the importance of network security audit technology in solving network security problems which around the network security model, the definition of security audit, some user behaviors that appeared in the process of network construction and application, the limitations of firewall technology and intrusion detection technology, etc.(3)In order to design the mobile Internet behavior audit system, this paper analyzed and compared three methods of machine learning, and chose semi-supervised machine learning method for processing training samples of the system. In order to improve the generalization capacity of machine learning, this paper further leads to the concept of ensemble learning. Through the analysis of the defects existing in the process of ensemble learning, the paper chose the selective ensemble learning method which is better than the ensemble learning method for the mobile Internet abnormal behavior detection model.(4)Semi-supervised machine learning and selective ensemble methods were used in the mobile Internet abnormal behavior detection model. In the process of detection, this paper gives a detailed definition of the concept and an integral structure of the model. Also, the paper described several algorithms used in the detection model in detail, including a simple ensemble algorithm based on clustering, a construction algorithm of the base classifier based on the mixing perturbation and the semi-supervised learning and selective ensemble algorithm of base classifier. In addition, the effectiveness of the method was verified by a simulation experiment.Finally, this paper designs and implements a mobile Internet behavior audit system, including data acquisition module, behavior audit module, Log alarm module and using MySQL to create the system database. Furthermore, through the final results of system testing shows that the system have good usability and stability.