Research On Cloud-Edge-End Collaborative Application Offloading Mechanism

With the rapid development of the Internet of Things technology and the formal commercialization of 5G,domestic and foreign standardization organizations have started to focus more on the edge computing field.The problem of secure access to billions of connected devices in cloud computing and the massive data problems can better solved in the edge computing mode,but the benefits of edge computing in processing applications with large resource requirements are very small.It needs to combine the strong resource advantages of the cloud.To this end,this thesis proposes a cloud-edge-end collaboration computing solution.This solution fully combines the advantages of sufficient resources in cloud computing and low latency of edge computing to rationally allocate computing resources and shorten the completion time of applications.The main research contents of this thesis are as follows:(1)Aiming at the problem of resource discovery in cloud-edge-end collaborative computing,this thesis proposes a new type of cloud-edge-end collaborative computing architecture.Based on this,a new type of resource discovery protocol is proposed to make full use of the gateway computing,storage function.The terminal initiates a resource request,and the gateway performs resource discovery.Unlike the traditional resource discovery protocol,this protocol can help the terminal shorten the cloud-edge resource discovery process,obtain real-time cloud-edge resources,and fully reduce the load of the core network.(2)Aiming at the problem of partition offloading of application components,this thesis proposes a method-level partition granularity,multi-offload destination dynamic application offloading algorithm.The algorithm firstly decouples the interdependence of application components,secondly estimates the calculation time and data transmission time of application components at various locations near the cloud,and builds a component cost relationship model.Finally,the topological ordering of application components without interdependencies generates multiple an application component call sequence set,using the greedy algorithm to determine the component uninstall location.The experimental results show that the cloud-edge-end collaborative application offloading algorithm proposed in this thesis has better time benefits than some traditional offloading algorithms in scenarios with many application components and large differences in computing requirements.(3)The forementioned resource discovery protocol and application partition algorithm,this thesis designs and implements a cloud-edge-end collaborative computing prototype system,develops functional modules for terminals,edge gateways,and cloud platforms,and tests the prototype through face recognition applications.Whether the system can successfully complete the calculation migration and shorten the completion time of the application.Test results show that the system can fully realize the expected function.