Design And Implementation Of Containerized Edge Computing Platform For Intelligent Video Surveillance Scenario

In recent years,the development of hardware and software technology of user terminal has greatly changed the shape of information and communication technology industry,which also makes the terminal produce a lot of data and requests.The centralized processing mode represented by cloud computing is becoming more and more difficult to cope with this situation,so offloading of computing in other parts of the service link,namely of edge computing,has become an option considered by many researchers and industry manufacturers.Video surveillance scenario is one of the most demanding fields for network,computing,storage,etc.It is a typical scenario of edge computing landing and also one of the most difficult scenarios.Based on K8 s and its derivative framework,which has become the de facto standard system for PaaS(Platform as a Service)construction,thesis explores the key requirements of edge computing application for PaaS service in intelligent video surveillance scenarios and carries out engineering implementation.PaaS is a development pattern that treats the running and development environment of applications as a service.For edge computing,there are two core differences from cloud computing PaaS:(1)The concept of computing resources has been extended.In addition to the general programmable computing resources managed in cloud computing PaaS,edge computing mainly adds terminal equipment as a data source,which needs to be incorporated into the computing resource management of the PaaS and provides service.(2)The communication environment has changed.Due to the actual network environment and geographical isolation of terminal devices,edge computing losts the assumption of stable and interworking network environment in cloud computing.PaaS of edge computing needs to consider how to provide basic services of cloud-edge collaboration.The work done in thesis being from the above two core differences,guided by the application requirements of a company's actual intelligent video surveillance system,solves the PaaS capability problems that the current K8 s and its derivative frameworks cannot solve:1.Solve the access and management of edge computing devices,especially camera devices,and abstract them into virtual computing resources of PaaS to provide data services.2.For the special communication environment of edge computing,the edge computing PaaS should provide a full range of cloud-edge collaboration basic services for container orchestration,configuration collaboration,and data collaboration,while the current framework only achieves the first two goals.Thesis designs and implements a configurable,high-performance and schedulable cloud-edge communication component to solve the cloud-edge data collaboration problem.In addition to the communication component for ordinary data,thesis also designs and implements communication components specially for video streaming.3.Solve the problem of how the edge computing PaaS provides suitable interfaces for applications to access the above device data services,cloud-edge communication services,and other platform basic services.Especially the solution mode of interface security problem,due to the special communication environment of edge computing,the focus is to avoid providing applications centralized security interface like in cloud computing PaaS-to avoid the situation that platform basic services need to query cloud services whether the edge computing application has the corresponding permissions,and performs high-latency cloud-edge communication,reducing the service quality.In the end,thesis completes the implementation of the platform interface,develops the intelligent video surveillance application,and completes the platform capability verification.