Research On Service Cache And Request Scheduling Strategies In Mobile Edge Computing

With the rapid development of mobile Internet and Internet of Things,more and more computation-intensive and time-delay sensitive applications emerge,and traditional cloud computing can no longer meet these needs.Mobile Edge Computing(MEC)provides users with efficient services by distributing resources at the edge of the network,while also reducing the latency associated with long backhaul transfers in cloud computing.Based on the characteristics of MEC distributed resource deployment,how to formulate effective resource allocation and request scheduling strategies to improve system performance has become the focus of current research.Most of the existing studies on computing offload assume that Edge Server(ES)can handle all types of user requests.It is ignored that ES can process the corresponding service request only when the data or library related to the service is cached.Because the storage capacity of ES is limited,all types of services cannot be cached.Therefore,the Service Cache and Request Scheduling(SCRS)problem must be considered.In this thesis,the SCRS problem is studied from two aspects: system delay and system utility.SCRS strategy is directly related to system delay optimization in cloud edge collaboration scenarios.In order to reduce system latency,ES provides services to users in a collaborative manner.Considering that ES needs to update the cache service type in real time to meet user requests.In this thesis,the SCRS problem with long-term service cache cost constraints is formalized as a mixed integer linear programming problem.Using Lyaplov optimization technique,the long-term optimization problem is transformed into real-time optimization problem,and then solved by dual decomposition algorithm.Through simulation experiments,it is verified that the proposed algorithm can approach the optimal solution under the constraint of long-term service cache cost.For the problem of system delay optimization in cloud edge collaboration scenario,SCRS strategy directly affects system effectiveness.The utility of ES is defined as the difference between benefits and costs.Considering that ES may belong to different network operators or private users,three modes of ES are defined: non-cooperative mode,full cooperative mode and incentive cooperative mode.In order to maximize system utility,the SCRS problem is formalized as a mixed integer linear programming problem in the fully cooperative mode,and a random-dependent rounding algorithm is proposed to optimize SCRS decisions for ES without violating the ES storage constraints.Subsequently,an incentive cooperation model is designed to encourage privately owned ES to join the cooperative system,and an alliance formation algorithm based on alliance game is developed,which can form a stable alliance for ES in the MEC network.The simulation results show that the random correlation rounding algorithm can greatly improve the system utility,and the alliance formation algorithm can effectively improve the ES utility.