Research On Service Placement And Request Scheduling In Edge Computing

With the development of the Internet of Things and wireless communication technologies,cloud computing has been unable to meet the low-latency requirements of some applications.And the massive amount of terminal data is handed over to the remote cloud for processing,which puts huge communication pressure on the core network.As a new computing paradigm,edge computing provides potential solutions to the above-mentioned challenges.Edge computing provides users with services by deploying storage,computing and network resources in a distributed manner.Based on the characteristics of distributed resource deployment,it has become a current research hotspot to find a reasonable resource allocation and request scheduling scheme for edge computing.The users offload tasks to edge nodes,which can reduce the energy consumption of terminal devices and shorten the processing time of tasks.The existing computing offloading work defaults that all types of requests can be processed on edge nodes.However,processing tasks on the edge nodes often depend on the services deployed by the nodes.Due to the limitation of storage capacity,the types of service installed on the edge nodes are limited.Therefore,it is necessary to consider request scheduling with service deployment This paper considers the service deployment and request scheduling problems in edge computing from the perspective of service provider revenue and service quality.In the scenario of multiple service providers(SPs),service deployment and request scheduling strategies are directly related to the revenue of SPs.In order to improve the service hit rate of edge nodes,nodes deployed by SPs provide services to users in a collaborative manner.Taking into account the change of service popularity,edge nodes need to constantly adjust the service to meet user requests.This paper formalizes the service deployment and request scheduling problem with long-term service replace-ment cost constraints as a mixed integer programming problem.Due to the lack of long-term prior knowledge,this problem cannot be solved offline.Based on Lyapunov optimization technology,the long-term optimization problem is converted into a series of real-time problems,and then a dual decomposition algorithm is used to iteratively solve the real-time problem.This paper analyzes the performance of online algorithm from a theoretical level,and the simulation results also show that the proposed algorithm increases the revenue by 25%.In the edge service system,different types of requests have different requirements for delay.Due to the limited resources of a single edge node,the system composes edge nodes in a certain area into a resource pool and provides services for users through unified management.The system expects to minimize the overtime loss of all tasks.In order to solve this problem,the paper proposes effective heuristic algorithms.Among them,the particle swarm algorithm reduces the complexity to search service deployment scheme,while the QoS-oriented request scheduling algorithm balances the load on nodes,and requests with high response time requirements are processed first in the nodes.The simulation results show that the proposed algorithm reduces the response time by 28%and the overtime loss by 40%,and improves the service satisfaction by 25%.