| With the development of mobile Internet services and applications,requirements for data rate and quality of service are highly increased.Mobile Edge Computing(MEC),as a key technology in the future communication technology,whose computing ability is strong because of rich computing resources in servers,the response time delay of MEC server to the user's computing task request and the possibility of network congestion between the transmission network and the core network are reduced by narrowing the geographical distance between the MEC server and the users.For users' mobile devices,in the limit of computing resources and battery power,offloading tasks to the MEC server will improve the energy efficiency of mobile devices.However,due to the selfish nature of users,the more computing resources are prompted,the more resources are wasted.In addition,user is not stationary and the coverage of the MEC server is limited,so offloading requests will be transmitted repeatedly if computing results isn't received in time after changing the location,which will cause computing resources wasted and energy efficiency of mobile devices reducing.The above two problems are more serious in resource-constrained scenarios,so it is worth better optimizing energy efficiency of mobile devices and better managing computing resources of servers in MEC for further study.In this thesis,from the perspective of improving the energy efficiency of mobile devices,the method of improving the energy efficiency of the mobile devices which competing for MEC servers with limited computing resources in a multi-user system is studied.In order to solve the problem of energy efficiency reducing and resources wasting caused by users' contention for resources and interference in multi-user system,multi-dimensional game algorithm for multi-user queue offloading(MQO)based on game theory(GT)is proposed.An upper limit on the computing resources of MEC server is set,so queuing will occur when MEC server is overloaded.The goal is to solve the problem of low energy efficiency of mobile devices in resource-limited multi-user systems.The players of the GT model in this algorithm are mobile users.The decision space is a two-dimensional space composed of offloading decisions and power control.The payoff function of player is energy efficiency and time delay function.The theory demonstrates that the Nash equilibrium of this model exists and is unique.Simulation results show that MQO,compared with the scheme which do not consider power control,can reduce energy consumption and time delay of mobile devices under the condition where the same task is completed and the number of offloading users increases,thus energy efficiency improved.In the scenario where the MEC server has limited coverage,this thesis aims to manage computing resources effectively in order to solve the problem that computing resources is wasted caused by user movement.Connectivity in this model has nothing to do with interference and competition in multi-user system and is closely related to geographical location.Single-user system is studied in order to better study the connectivity between user and MEC server.To solve the connectivity problem between a single mobile device and MEC server,computational resource management algorithm based on extended kalman filter(CRMAEKF)is proposed,whose main steps are prediction and measurement feedback.Through finite iterations and parameters' modified,the optimal prediction position of the mobile user who is doing non-linear motion is obtained.Finally,the MEC server in the area that can continuously provide service to moving user based on the position is selected.Computing resources can be fully utilized by reasonable allocation of MEC server.Compared with the no prediction server selection scheme,simulation results show that,the success rate of the connectivity between the selected MEC server and the user is significantly improved by CRMAEKF. |
PDF Full Text Request