Research On The Joint Optimization Of Service Replica-Tion And Request Offloading For Edge Computing

In recent years,the Internet of Things(IoT)has developed rapidly,and the demand for services such as time-sensitive applications such as video streaming,traffic information,personalized multimedia,and data sharing has increased rapidly.Edge Computing emerged as the times required,and its core idea is to offload some requests to network edge nodes for processing instead of offloading them to cloud servers for processing.However,in the mobile edge computing architecture,the computing and storage resources of edge servers are usually limited.When the number of IoT devices connected to the edge computing architecture increases,the rational allocation of resources is extremely important.To this end,starting from the computing and storage resources provided by mobile edge devices,this paper studies the problem of joint service replication and request offloading.The main contents are summarized as follows:(1)In the multi-edge single-user scenario,a single service replication cannot meet the request offloading requirements of users in different topological locations,a redundant replication combined service replication and request offloading strategy is proposed to minimize user request delay.This paper first analyzes this problem as a Pareto optimization problem.Considering the heterogeneity of edge servers on edge Computing,considering the necessity of redundant service replication,the Pareto optimization problem is planned as a joint multi-service replication and Problems requesting offloading.Then,a multi-service replica replication model for edge computing is established to solve the problem of how to perform service replication on edge servers and how to offload requests on edge servers with replications.Simulation results show that the algorithm which is proposed in this paper can effectively reduce the overall delay.And for larger problem instances,only a short execution time is required to make decisions.(2)In the multi-user multi-edge scenario,considering the limited resources of the Mobile Edge Computing(MEC)server and the complex MEC network,the problem of joint service replication and user request offloading to minimize system energy consumption is solved.Each edge server can process multiple user requests,which does not mean that all requests can be processed by the same edge server in the MEC scenario.This paper proposes a divide-and-conquer heuristic approach to study service replication and user request offloading strategies in multi-user multi-edge scenarios.This paper considers the two possibilities of processing user requests locally and offloading them to edge servers.First,the original problem is divided into several balanced sub-problems by using the graph partition method,and then an improved particle swarm optimization(IPSO).The simulation results show that,compared with other algorithms,the algorithm can significantly reduce the total energy consumption of the system.(3)In the multi-user multi-edge scenario,the problem of service replication and user request offloading to maximize the benefits of edge service providers is studied in view of the current situation that users may hide real offers.First,under the conditions of limited MEC server energy and user request deadline constraints,the problem of j oint service replication and user request offloading is proposed to maximize the profit of edge service providers.Secondly,according to the correlation between user quotation and MEC service provider revenue,the problem of service replication and user request offloading is reduced to a Pareto optimization problem.Finally,the deep Q-network(DQN)algorithm is used to achieve the goal of maximizing the service provider’s revenue by meeting the MEC server energy and user request delay.The simulation results show that the algorithm proposed in this paper can effectively improve the efficiency of the proposed algorithm on the basis of ensuring that the user’s request delay does not exceed the maximum deadline and the service replication energy consumption MEC server’s maximum energy reserve without relying on the prior knowledge of the network state.Service provider revenue.The simulation results show that the algorithm proposed in this paper can effectively improve the revenue of service providers on the premise of not relying on the prior knowledge of network state,ensuring that the user request delay does not exceed the maximum deadline,service replication energy consumption and the maximum energy reserve of MEC server...