Research On Server Resource Deployment And Service Offloading Mechanism For Intelligent Transportation Service

Multi access edge computing,as an emerging technology to improve business processing efficiency and relieve the pressure of terminal computing,has been applied in more and more fields.The edge computing technology provides the computing capacity and storage capacity of the cloud computing center at the edge of the network near the terminal device,so that the processing requirements of delay sensitive services can be quickly responded to.When the battery energy of the terminal device is insufficient,the services with large energy consumption will be unloaded to the edge computing server for processing.Intelligent transportation uses technologies such as the Internet of Things and artificial intelligence to efficiently process and analyze urban traffic data,and combines edge computing network to build an edge-end processing system to improve the management efficiency of traffic business.In order to make the vehicle business more secure and efficient,how to reasonably deploy the resources of the edge computing server and set the matching scheme for business offloading is an urgent problem to be solved.At present,the research on resource deployment focuses on the characteristics of resources and lacks the analysis of the business requirements themselves,such as the analysis of the similarity of business and service resources and the analysis of business density.In terms of service offloading,some documents only consider single decision of delay or energy consumption.Some researches lack preprocessing processes such as classification and prediction of services.Aiming at the problems of resource deployment and service offloading in intelligent transportation scenarios,this thesis proposes a server resource deployment and service offloading mechanism for intelligent transportation services:(1)This thesis proposes a resource deployment mechanism based on characteristics of business requirements and equipment service capabilities.First,the system model is established according to the business scenario,including the business demand model,edge computing server model,similarity matching model,etc.The optimization goal consists of two indicators,namely,resource deployment cost and business processing time.By analyzing factors such as the matching degree between the business and the edge computing server,and the spatio-temporal characteristics of the business,resources are deployed according to the similarity and business density,so as to improve the efficiency of resource utilization and vehicle business processing.This thesis uses the idea of differential evolution algorithm to improve the ant colony algorithm.This algorithm can effectively solve the optimal deployment scheme.The experimental results show that the heuristic algorithm is improved to reduce the occurrence of local optimum and accelerate the convergence speed of the algorithm.At the same time,the resource deployment scheme proposed in this thesis can reduce the cost of resource deployment and improve the quality of the optimal solution.(2)This thesis proposes an offloading mechanism based on traffic classification and traffic volume prediction.Before making the offloading decision,the business requirements should be preprocessed first.The KMeans algorithm is used to classify the business demand according to the size of the data volume and the number of resources required.Then the future data volume is further predicted according to the classification results.In this thesis,the time series data is predicted by using the gated recurrent unit with good prediction effect and simple structure.According to the results of business preprocessing,adjust the server resources,and then make the offloading decision.Determine whether to offload the service according to the delay sensitivity of the service and the scarcity of service resources.For the services processed in the local device,the processing cost is minimized by dynamically adjusting the CPU frequency of the terminal device.For the services to be offloaded,the improved Kuhn-Munkres algorithm is used to plan the offloading scheme to ensure fairness for users and minimize the offloading delay and energy consumption.The experimental results show that the preprocessing scheme of business requirements proposed in this thesis can reduce the cost of resources.The offload matching scheme can minimize the offload cost and balance the allocation of computing resources.