Research And Implementation Of Service Management And Energy Consumption Prediction Mechanism Of Edge Computing

With the emergence of various mobile devices and devices of Internet of Things,new types of services and mobile applications that utilize machine learning and augmented reality technologies are emerging.These services usually require more computing resources on the cloud,lower latency and higher reliability,so edge computing technology has emerged.At present,edge computing technology has been applied in many scenarios.Multi-access edge computing(MEC)technology promotes the application of edge computing from the original mobile cellular network to non-3GPP(such as WiFi and fixed network)wireless access methods,which further promotes the popularization and Implement of edge computing technology.Although MEC technology can provide the complete task distribution models and application platforms,the MEC network still has to deal with many problems on its own.Therefore,the related issues of service management and energy consumption prediction mechanism are studied.The specific research work includes the following four parts:(1)Aiming at the current status of dynamic migration services in ultra-dense MEC networks,a hierarchical MEC architecture is proposed.This architecture integrates the MEC server cluster and base station in the radio access network(RAN)and deploys it on the user plane interface.In addition,a chain management with a value adjustment mechanism(CMVAM)is proposed.Based on the introduction of a credit system,this mechanism uses a solution similar to the multi-box packaging problem to solve the problem of migration service selection.Experimental results show that compared with other chain management mechanisms,this mechanism can more efficiently adjust the scheduling problem of dynamic service migration in ultra-dense MEC networks.(2)Aiming at the current status of the service evaluation mechanism in the MEC network,a migration effect evaluation(MEEQ)algorithm based on the perception of quality of service is proposed.The algorithm subdivides the user's perception domains,gives different evaluation indicators for services in different perception domains,and configures independent weights for different services.The experimental results show that compared with other evaluation algorithms,this algorithm can reflect the service quality of different services in more detail and fully.(3)Aiming at the current situation of research on energy consumption of edge nodes,a new energy consumption model for edge nodes is established.The energy consumption framework of edge nodes is first proposed,and then the energy consumption modeling process of edge nodes is gradually analyzed.The energy consumption models from the real-time and long-term perspectives are finally established.The experimental results show that this model has higher accuracy than other energy consumption models.(4)Using the Cloud Sim framework as the server to build a simulation platform for energy consumption and evaluation of the server cluster.The simulation platform is based on the MEC architecture and service evaluation algorithm proposed,and uses the energy consumption model established to simulate.Starting from five aspects: the complexity of the energy consumption structure of the server cluster,the applicability of server clusters with different structures,the convenience of user-friendly use,the accuracy of energy consumption simulation results and the degree of reflection of the evaluation results,it provides users with freely set up server clusters,energy consumption simulation of server clusters,and service evaluation of server clusters to meet the needs of people in related fields for such products.