| With the development of mobile communication technology into the fifth generation,new scenarios and applications such as virtual/augmented reality,autonomous driving cars,Tactile Internet and Internet of Things continue to emerge,leading to a significant increase in human society’s demand for communication,computing and other resources,and more stringent requirements on time delay.In order to meet the requirements of the above scenarios and applications,the academic circles put forward Mobile Edge Computing(MEC)technology,and continue to promote its application.However,the security of MEC hinders the development and application of MEC technology.Edge nodes are vulnerable and complex,which makes it difficult to deploy traditional security mechanism in MEC network.The access rights to user’s large amounts of sensitive information make edge nodes more likely to attract the attention of external malicious attackers.Therefore,the research and application in the field of security has very important academic value and practical significance,and has become one of the research hotspots in recent years.The security of edge nodes in MEC network is an important prerequisite to ensure the security of user data.Users have different requirements for security when using different applications,so it is necessary to select relatively safe edge nodes to ensure data security and protect privacy when the user carries out the load offloading of the computing task.In addition,in the MEC network,the logs of devices and applications will continuously record the network operation status.How to efficiently manage and analyze the complicated log data and find unexpected failures and security threats in time as well as evaluate the network security situation also needs in-depth research.The main work of this paper is as follows:First,in the MEC network scenario of cloud-edge collaboration,an offloading algorithm based on the credibility of edge nodes is proposed.Firstly,the security requirements of the user application and the reliability of the edge computing node are considered comprehensively,the historical data of users and edge computing nodes are preprocessed,the Support Vector Machine(SVM)multi-classification algorithm is used to train the offloading decision model.Then the corresponding classification decision function is used to select the edge computing node which is relatively reliable to the current user for offloading.The simulation results show that the trusted node offloading algorithm proposed in this paper takes into account both the node credibility and the offloading cost to a certain extent when the task is offloaded in the MEC network.Second,an MEC network security platform based on log analysis is built on Nvidia Edge Computing Platform.Firstly,the overall demand of the platform is analyzed according to its work content.Secondly,the overall design of the platform is discussed in detail,including the business process and module division of the platform.Thirdly,according to the target function of each module and the relationship between each other,the architecture of each module and the technology and framework used are designed and introduced.Finally,based on the previous research,each module is implemented.The platform built in this paper can effectively collect,process,analyze,store,retrieve and display all kinds of logs in MEC network,effectively improve the efficiency of network operation and maintenance,discover network security threats,reduce the cost of network maintenance,and provide a strong support for evaluating the security state of edge computing nodes. |
