Research On The Strategy Of Task Offloading And Server Sleep In Mobile Edge Computing

With the rapid development and integration of mobile communication technology and Internet of things(Io T)technology,a large number of Io T devices are deployed at the edge of mobile communication networks.However,most Io T devices have the problem of insufficient computing power.To solve such problems,mobile edge computing(MEC)has emerged.At the same time,to better provide differentiated services for these devices and improve the efficiency of the MEC system,heterogeneous network MEC(H-MEC)for heterogeneous networks has become a current important topic.However,the H-MEC system still has the problems of excessive energy consumption caused by massive deployments of servers and insufficient user equipment(UE)computing power and energy supply.As a result,this thesis discusses from the two aspects as follows.Aiming at the issue of excessive energy consumption caused by a large number of servers deployed in the H-MEC system,this thesis proposes a collaborative computing strategy with multiple sleep states of servers.First,a delay energy consumption model is constructed according to the sleep state of the server at different depths.Then,a model is established for the number of tasks arriving on the server non-uniformly,and a task migration strategy is proposed to model the problem of minimizing system energy consumption as a mixed integer nonlinear programming(MINLP)problem.However,the scale of the problem is too large.This article uses the slot division method to transform it into a small-scale MINLP problem and uses the Fmincon optimizer to solve it.According to the correlation between adjacent sub-problems,the solution strategy is updated by useing the migration update algorithm to ensure that the follow-up problems can be executed correctly.Finally,the sleep decision algorithm is used to determine which sleep mode the server enters.The simulation results show that the proposed algorithm can effectively reduce the energy consumption of the system.Aiming at the problem of insufficient UE computing power and continuous energy supply,this thesis proposes a task offloading strategy based on UE sleep.According to the non-uniform generation of tasks on the UE,the task offloading problem is constructed as a Markov decision process.To minimize the long-term average cost of the system,a task offloading algorithm based on asynchronous deep Q network(ADQN)is proposed.To further reduce the cost of the system,this thesis establishes a dynamic frequency adjustment sub-problem of the server to solve it.Simulation shows that the proposed algorithm can obtain the optimal task offloading strategy and computing resource allocation strategy,and significantly reduce the cost of the system.