Research On Computation Offloading Strategy For Mobile Edge Computing Network

With the decreasing of manufacturing cost,smart mobile devices have been widely used,which promotes the development of mobile Internet.On the other hand,the disposing of mobile Internet leads to the growth of mobile applications industry so that various mobile applications are emerging in the market.Now,mobile applications can provide diverse services,and their functions are more and more powerful.It may need excessive resources if running these applications.However,smart mobile devices are resources-constrained.It may cause the high energy consumption and the long task execution delay when they execute computation-intensive applications,thus resulting poor user service experience.Computation offloading is an effective way to solve this problem.But,the traditional remote cloud computation offloading leads to large service delay.Therefore,the mobile edge computing(MEC)and the deploying architecture of mobile edge computing network are proposed,which provide a new approach to actualize the computation offloading.In the mobile edge computing network,MEC servers provide rich computing resources and storage resources nearby mobile users,and the smart device with adequate and idle resources can provide assisted computing services through Device-to-Device(D2D)link.Accordingly,with implementing computation offloading in mobile edge computing network,it can decrease the service delay,compensate the shortage of resources at the mobile user end,and further improve the quality of users' experience.On the research problem of computation offloading strategy for mobile edge computing network,the hybrid computation offloading scenario is current tendency.The main works of this thesis focus on this research direction.It contains two areas as following:(1)It requires task data transmission during the process of computation offloading.Therefore,cellular transmission interference and D2 D transmission interference among mobile users affect the computation offloading strategies in hybrid computation offloading scenario.Accordingly,in this thesis,the efficient computation offloading scheme in a hybrid computation offloading scenario is investigated.Specifically,mobile users can choose to offload its tasks to either the MEC server or a near distributed computing node(DCN),or execute tasks itself.Due to the offloading decisions of different users affect the task execution delay and energy consumption of each other,firstly,the decision-making problem is considered as a sequential game.Secondly,the existence of Nash equilibrium is proved so that it can be converge to stable state.Then,an multi-user hybrid computation offloading scheme is proposed to achieve Nash equilibrium.Finally,the performance of the proposed scheme is verified by simulation.And the simulation results show that the proposed scheme can promise in processing delay-sensitive computation tasks with a constrained energy consumption.(2)Actually,the generation of computation tasks for mobile users is random.In the hybrid computation offloading scenario,the efficient computation offloading strategy based on random task model is investigated.In this thesis,the computation task generation of mobile users is considered as a Poisson process.Thus,the execution process of mobile equipment,DCN equipment or MEC server is constructed as a queuing model.Then,the optimization problem is set up aiming at minimizing the cost of task execution.And a hybrid computation offloading decision-making algorithm based on the exterior penalty function method is proposed to solve the problem.It can obtain the optimal computation offloading strategy through the proposed algorithm.Finally,the validity of the proposed algorithm is verified by simulation.The simulation results show that the proposed algorithm has significant performance gains in both reducing delay and saving energy.And it shows great adaptability to different requirements of different mobile users.