Research On Task Offloading Method Of Moving Edge Computing D2D Communication

With the explosion of mobile devices,Io T devices,self-driving cars and other mobile connected devices,they use more traffic data than currently available network bandwidth can meet.The use of central cloud technology to process large amounts of data was briefly considered an effective solution.However,cloud computing has obvious disadvantages in the process of solving terminal computing demand: high latency,high energy consumption and high network bandwidth occupation.As a new data processing mode,Mobile Edge Computing sets several edge servers scattered around to relieve the computing pressure on the central cloud.MEC can be used to address the increasing computing needs of mobile applications on mobile devices.However,due to the limited size of each edge server,its own computing capacity is limited,and it is also unable to unload massive data to the mobile edge server.In addition,device-to-device(D2D)communication is a key technology in mobile edge computing.It is expected to improve spectrum efficiency by reusing the same radio resources as cellular users,while saving wireless bandwidth.However,because mobile users are in multiple network signal categories,there will be serious interference problems,so reasonable allocation of resources is very important.Therefore,this paper edge for mobile computing tasks under offloading and D2 D resource allocation research,the main innovation points are as follows: 1)in the existing based on discrete action,on the basis of space to make a decision,this paper proposes an algorithm based on continuous motion space of the candidate networks,in order to obtain better power control of the local task execution and offloading.2)Based on the multi-user MEC system,the dynamic offloading strategy is independently learned for each mobile user with task random arrival and time-varying wireless channels.Based on this strategy,ECOO(Edge Computing Optimize Offloading)algorithm is implemented to reduce power consumption,Computing cost and delay.3)Let each D2 D pair select the D2 D mode according to the actual link quality to reduce the impact of unreliable D2 D links.In this paper,a joint problem of transmission mode selection,physical resource block(PRB)allocation and power control for D2 D communication is proposed under the premise of satisfying D2D’s requirements on time delay and reliability.4)In order to maximize the system throughput,a scheme combining the three algorithms is proposed.In this scheme,a lightweight heuristic algorithm is used to achieve the optimal selection of patterns,and then the Hungarian algorithm is combined with the heuristic algorithm to achieve the optimal solution of channel and physical resource block allocation.Finally,DQN algorithm is used to allocate power and make adaptive decision according to local observation.5)A new method is proposed to bridge channel feedback and preencoder feedback by combining design feedback and precoding strategy under the framework of team decision.Users and base stations minimize the common mean square error(MSE)measure based on their individual observations of the imperfect global CSI(channel state information).Through a large number of experiments and simulations,it is proved that the work done in this paper can effectively reduce the system delay and energy consumption,and reduce communication interference to improve the throughput.