Research On Offloading Algorithm And Strategy Of Mobile Edge Computing Based On Position Prediction

With the rapid development and popularization of 5G(The Fifth Generation Mobile Communication System)technology,many computation-intensive mobile applications not only provide users with high-quality service experience,but also bring great challenges to mobile terminals with limited computing resources.Mobile Edge Computing(MEC)deploys servers at the edge of networks to provide computing services for terminals,meeting the requirements of low latency and power consumption,and effectively alleviating computing pressure on mobile terminals.Due to the mobility of users,the limited coverage and the constrained computing resources of MEC servers,offloading with large number of tasks will bring high latency.To deal with the problem,a collaborative offloading algorithm of MEC severs based on location prediction is proposed in this thesis,which allows multiple servers to work cooperatively for providing services of computation offloading.The user’s future position is predicted with extended Kalman filter(EKF),and then the appropriate servers are chosen to handle the offloading task.Based on the prediction,the divided task is assigned to multiple servers in order to minimize the weighted sum of delay and energy consumption,then a utility function is given and optimized by the simplex method.Simulation results show that the proposed algorithm has better performance in the communication delay and energy consumption rather than the existing methods.And the success rate of connecting between the user and the servers has also been improved.Aiming at the limited computing resources and load imbalance of MEC servers,the current load status and the location of MEC servers are considered comprehensively in this thesis.A load balancing-oriented MEC task offloading strategy is proposed,in which part of the workload of overloaded MEC server is transferred to the adjacent MEC server with relatively light load.This strategy is designed to balance the load of MEC servers,reduce the processing time of users’ offloading tasks,and make full use of the computing resources in MEC servers.Simulation results show that the proposed strategy has lower task processing delay and more balanced load state of MEC servers compared with the traditional strategies.