Dynamic Mobility-Aware Offloading And Resource Allocation In Mobile Edge Computing

The popularity of user equipment and the diversification of mobile applications have brought about Mobile Cloud Computing(MCC),while the higher latency and energy requirements of tasks lead to Mobile Edge Computing(MEC).Mobile Edge Computing brings computation closer to the user equipment(UE)and reduces the latency as well as energy consumption for computation offloading.However,in order to solve the challenge of the explosive growth of data traffic in the fifth generation mobile communication network(5G),it is necessary to increase data traffic and enhance user experience.Ultra-dense network(UDN)is one of the main technologies of the fifth generation mobile communication network where low power base stations are deployed intensively.With a frequent handover between base stations,traditional offloading algorithms and mobility management(MM)approaches are not sufficient anymore.It becomes far from optimal to execute offloading algorithms only at the beginning of a task when UE moves frequently or in the environment of UDN.In this paper,a dynamic mobility-aware partial offloading and resource allocation approach is proposed.This paper first establishes the MEC offloading model and formally describes the MEC offloading decision problem.In this paper,linear functions are used to express the changes in delay and energy with the increase of local computing data amount,and function diagrams are drawn.From the graph,the optimal value of offloading strategy under different conditions can be obtained intuitively.At the same time,this paper formally describes the problem to solve,and describes the goal of minimizing energy consumption while meeting the delay in a more rigorous language,so as to find a way for the computer to understand and compute the optimal solution.This paper abstractly expresses the architecture model of mobile edge computing.The architecture is divided into three layers(data layer,control layer,software definition service)and three roles(user equipment,micro base station,controller).The controller is responsible for collecting information of base station and computing resources as well as making offloading decisions;base station is the place where the user equipment directly connect to and offload tasks;the user equipment manages its own tasks.The control layer is the layer where the user equipment communicates with the controller to make offloading decisions,while the data layer is the layer where the user equipment offloads data to the base station for task calculation.The separation of the control layer and the data layer achieves the purpose of centralized control and reduces the pressure of the controller.Mobility prediction is integrated into the offloading strategy.Unlike the common offloading decision-making method,this paper combines the offloading strategy with the mobility management module.The method proposed in this paper will assign the amount of data to the location with better network conditions in the future as far as possible according to user mobility prediction.That is to say,this paper predicts user movement in a short future time interval and assigns tasks to be processed for this future period.This allows tasks to choose a more suitable place for offloading.By combining mobility prediction with offloading decision and resource allocation,this paper can consider the changes in network conditions when making decisions,achieving the goal of minimizing energy consumption while meeting the delay constraints.Due to the change of the network environment and the inaccuracy of mobility prediction,the decision-making result of this paper is only a near optimal offloading strategy in the next short period of time.However,due to the dynamic adjustment of the offloading strategy in the whole process,the approach in this paper can achieve a good effect on the whole task process.In order to verify the effectiveness of our approach,this paper compares this ap-proach with two traditional methods:partial offloading strategy with mobility man-agement(POwMM)and dynamic partial offloading strategy with mobility management(DPOwMM).This paper verifies the feasibility and validity of the approach in this paper by comparing the delay requirement satisfaction rate and the energy saving ratio under different parameters in the situation when the user equipment moves.The results show that in the case of non-extreme motion,the approach in this paper can achieve better delay requirement satisfaction rate and save more energy than the compared method,especially when the time requirement is not strict and the accuracy of mobility prediction is high.