Task Offloading In Mobile Edge Computing

In recent years,mobile networks are moving towards the 5th generation mobile networks at a high speed.The types and number of devices connected to mobile networks have grown rapidly.New mobile applications are increasingly demanding computing resources and low latency.These development trends place higher requirements on the current mobile network architecture and cloud infrastructure,spawning the mobile edge computing（Mobile Edge Computing,MEC）model^[1].In the MEC model,user tasks are deployed on mobile edge servers,enabling user equipment to obtain longer battery life,lower processing load,shorter network delays,and reducing network congestion.During the deployment of mobile user tasks,latency and energy consumption are two key factors that affect system performance.Therefore,for different network architectures and different application scenarios,this thesis studies and explores the deployment strategy of mobile user tasks between mobile edge servers,including the following:1)Deployment strategy optimization under multi-user-single-macro-base-station architecture.First,for the users’computing offload scenario of in the macro base station,study the deployment method of computing tasks to minimize the energy consumption of the MEC server next to the macro base station.This thesis considers the three resource constraints of the CPU,memory and storage of the MEC server,and designs a computational deployment algorithm FMLP that minimizes server energy consumption.Then,this thesis studies the performance optimization methods for multi-service deployment scenarios of macro base stations and users.In this scenario,users need to deploy computing tasks and specific application service requirements on the MEC server at the same time,and the mobile application operator loads various application services on different servers of the MEC nodes.This thesis needs to minimize the transmission delay of user application requests in the network under the constraints of computing offload resources.Therefore,this thesis proposes an allocation algorithm FMMLP that minimizes service transmission delay.In the simulation experiments of the two algorithms,FMLP and FMMLP have achieved the results of minimizing energy consumption and minimizing delay in various scenarios.2)Deployment strategy optimization under multi-user-multi-small-cell architecture.For the task offloading scenario composed of small base stations and nearby users,considering the change of user’s location and the change of connectable base stations,the scenario is modeled as two stages of network access and user task offloading.An iterative algorithm OMMA is designed,which satisfies the constraints of base station resources and server resources,and minimizes the total energy consumption of all MEC servers as the highest priority optimization goal.Then iteratively finds the solution with the minimum delay in the solution set that minimizes energy consumption.In the comparison experiment with the OLSA algorithm of Gao[2],the OMMA algorithm can find the solution with the lowest total delay among all solutions that minimize energy consumption,and the energy consumption is reduced by 16.47%～56.08%compared with the OLSA algorithm.