Service Placement And Request Scheduling Strategy In Edge Computing

In the edge computing mode,edge devices with storage capacity and computing power can preset specific service programs and process user service requests to reduce service latency and reduce the traffic burden on the core network.Due to the limited resources of edge devices,all application services cannot be placed,so all service requests of users cannot be satisfied.Finding the optimal service placement and request scheduling scheme for user requests has become a hot research topic.This thesis studies from two aspects: deadline constraint and multiple Service-Provider(SP)cost constraint.The main work is summarized as follows:(1)Deadline-constrained service placement and request scheduling strategy.The strategy combines the deadline to build a task completion revenue model,and then proposes a service placement method based on particle swarm algorithm,combined with a heuristic request allocation algorithm,to find a reasonable service placement and request allocation scheme through iteration.On this basis,an online task scheduling algorithm is designed to maximize the benefits of task completion by prioritizing tasks with high response delay requirements.The simulation results show that the algorithm proposed in this thesis can preferentially cache and process application services and tasks with high response delay requirements,thereby improving the task completion rate and task completion revenue.(2)Multi-SP cost-constrained service placement and request scheduling strategy.This strategy combines edge layer resource consumption and service cache update cost to build a service provider operating cost model,and builds a method to solve the service placement and request scheduling problem with long-term cost constraints of being served.The long-term optimization problem is transformed into a series of single-slot optimization problems by Lyapunov optimization,and a single-slot service placement and request scheduling algorithm is proposed.The simulation results show that,compared with the baseline algorithm,the algorithm proposed in this thesis can obtain the service placement and request scheduling scheme with the minimum delay under cost constraints.