Research On The Construction And Deployment Of Edge Computing Gateway Devices

With the emergence of the new paradigm of edge computing,we can put local data center resources on the edge of the Internet of things to improve the transmission efficiency of these terminals to the edge services by geographic location closer to mobile devices,sensors,and all kinds of emerging devices for Internet of things to ensure end to end milliseconds low level delay.Edge computing is an open computing platform that combines network,computing,storage,and application capabilities,and is closer to the source distance of objects or data in the network.ZTE,HUAWEI,Intel and other Internet Co have already proposed a platform or model for providing edge computing services,and now researchers have also done some deployment algorithm research and Optimization on the allocation of virtual resources in edge computing and the unload of computing tasks.Through the mastery of the above research situation,this paper combines the existing research results to further study the problem of edge computing gateway equipment.The main contents of this paper are as follows:First,this thesis introduces how to build a simple gateway device to provide edge computing services,and to build a raspberry pie into a gateway device,combining the features of the raspberry pie.By establishing a wireless routing module,a data analysis module,and a device management module to the raspberry pie,a single independent establishment of a gateway device that provides edge computing services is completed.The gateway device designed in this paper not only has a narrow sense of router,but also takes on the task of providing service for the edge computing,and meets the requirements of the future development of the IOT gateway.Secondly,it mainly combines the research status of edge equipment deployment in edge computing model.By constructing the edge computing model to the existing network,the Gauss model component is used to fit all terminal equipment in the area.By adding some limited indexes,the probability distribution is solved by solving the maximum likelihood estimation.The model is used to determine the number of Gauss components with Bayesian information criterion,thus determining the final distribution model.In the process of solving the iteration,the algorithm completes the clustering of the sample nodes and the selection of the cluster center.Finally,the advantages of the deployment method are illustrated by the experiment.