Research On Computation Offloading And Task Caching Strategy For Mobile Edge Computing Network

With the rapid proliferation of mobile devices and the development of wireless communication technology,more and more novel and complex applications have emerged,such as face recognition,augmented reality(AR)and interactive games,which require a lot of computing resources.This poses a huge challenge to mobile devices with limited resources.Computing offloading is a mainstream technology to reduce task execution delay and save energy for mobile users.Among the current computing offloading methods,mobile edge computing(MEC)is a promising technology,which has caused extensive discussion in industry and academia in recent years.The concept of MEC was first proposed by the European Telecommunications Standards Institute(ETSI)in 2014 to provides information technology and cloud computing functions in wireless access networks near mobile users.However,the existing computing offloading technology requires data transmission between the mobile user and the MEC server.Because the energy consumption and delay caused by data transmission occupy the main part of the overall computing cost of mobile users,it is difficult to meet the computing demands of users in some computing-intensive applications with huge data volumes.Motivated by this,this thesis improves the existing MEC technology by introducing task code caching at the MEC server.Specifically,it allows users to cache user tasks at the MEC server proactively,and allows users to offload tasks to the MEC server for execution as in the traditional edge computing architecture.This thesis mainly completed the following two aspects of work.(1)This thesis studied the joint optimization problem of computing resource allocation,task caching and computation offloading in MEC-enabled mobile wireless communication networks.By modeling the computational offloading and task caching problems as mixed integer non-linear programming problems,the aim is to minimize the weighted sum of the task execution delay and energy consumption of all users.In this thesis,an alternative optimization framework is used to solve the optimal computing resource allocation,optimal task caching strategy and optimal computation offloading strategy through three steps,and establish the upper bound of the optimal performance of the system design.Firstly,the optimal resource allocation strategy is addressed by using convex optimization method.Secondly,the optimization problem of the summation of the offloading strategy and the task caching scheme is converted into an equivalent linear programming problem,which are solved by Karmarkar algorithm in polynomial time.Finally,the optimal task offloading strategy is addressed,and the optimal task caching scheme is solved based on the optimal summation of the task caching strategy and the computation offloading decision.The simulation results show that the proposed scheme can significantly reduce the system cost and meet the user's requirements for task execution delay and energy consumption.(2)This thesis studied task caching-based MEC computation offloading,which jointly optimize the computation offloading strategy,the task cache strategy of the MEC server,and the wireless channel allocation strategy.This thesis integrated the task caching mechanism into the traditional edge computing offloading technology,and proposed a new computing architecture.In the proposed computing architecture,the MEC server can pre-cache part of the user's computing tasks.When users need executor tasks,they can complete their own computational tasks through three methods: local computing,computation offloading,and requesting the MEC server to directly execute cached tasks.The aim is to minimize the total energy consumption of all users while satisfying the task execution delay constraints.By modeling the problem as an integer programming,this thesis takes two steps to find the optimal solution.Firstly,the task offloading strategy is addressed by using game theory.Secondly,the optimal task cache strategy is solved by dynamic programming.The simulation results show that compared with the other two benchmark schemes,the proposed scheme can effectively reduce the system cost.