Research And Implementation Of Business Risk Prediction In Mobile Edge Environment

With the rapid development of information technology,more and more businesses in today's society can be completed through the Internet,and the use of the Internet and the business data it generates are constantly improving.Most of the existing risk prediction systems rely on the central server to complete,but this may result in a decrease in processing efficiency or a deviation in results due to differences in the location,network conditions,and load conditions of the business personnel.Most of the existing risk prediction systems rely on the central server to complete data processing and risk prediction,but they do not consider the mobile edge environment and data overload.On the one hand,if only relying on the central server to provide prediction function,the prediction quality will be affected by the distance between users and the central server and the different network conditions.On the other hand,if a large number of users transmit data to the central server through mobile devices,the load of network bandwidth will increase,and the network delay will also increase,which will affect the prediction work.To solve the above two problems,this paper proposes a business risk prediction in edge computing environment.The main contents of the work are as follows:Firstly,according to the characteristics of risk data,parallel processing technology of ETL and big data is used for data processing,machine learning algorithm is used to establish business risk prediction model,and processed data is used to train model.Then,because edge devices are closer to users and have certain computing and resource storage capabilities,this paper uses edge devices to assist the central server in risk prediction under mobile edge environment.At the same time,because computational unloading mainly migrates heavy computational tasks to edge servers to improve the performance of mobile services,it can not solve the problem of data overload in central servers.Therefore,this paper proposes the concept of data unloading,and designs the corresponding data unloading platform framework,and uses the edge device to assist the central server to alleviate the data overload problem to ensure the efficiency of forecasting work.Then,according to the platform framework of data uninstallation,corresponding data uninstallation strategies and algorithms are designed.The specific strategies are as follows:1)An edge device cluster is built at the edge,and the cluster is used to process the edge data.2)In the data processing phase,the edge device cluster is used to take care of the data processing phase to assist the central server in sharing computing and data pressure.3)In the risk prediction stage,Control the data sharing according to the distribution coefficient and the load of the central server,where the coefficient can be set by the user;then,at the central server,the pressure is detected in real time,and the data sharing is controlled according to the pressure;Edge clusters set data distribution policies to ensure the processing efficiency of the central server and edge device cluster.Finally,this article takes the customs historical customs declaration data as an example.The data is desensitized and deformed by sensitive data to meet the customs-related data security regulations at all levels.Design and implement a risk prediction system,and experiment with the data unloading strategy proposed in this paper.The experimental results show that the strategy can achieve the unloading effect and reduce the data pressure of the central server.