Research On Energy Efficient Computing Offloading Strategy In UAV-Edge-Cloud Computing Environment

As the key technology foundation of 5G,mobile edge computing(MEC)provides users with computing,communication and storage services at the edge of the network by sinking the service node closer to the user,so as to reduce the delay and energy consumption caused by processing tasks.Considering the flexibility and easy deployment of UAV,it can be used as a mobile edge server to provide effective communication coverage for terminal equipment.In addition,it can also be used as an energy supply platform to provide energy for task offloading for devices by using wireless charging technology.However,problems such as signal blockage and shadows caused by the environment,and limited resources carried by UAVs can all prevent successful offloading tasks to UAVs.In addition,Io T terminal equipment in remote areas also has the problem of inconvenient power supply.Therefore,this paper studies the problems of UAV energy supply and task offloading.The research work of this paper is summarized as follows:1)Aiming at the inconvenient power supply of terminal equipment,a wireless power supply MEC network based on a single UAV is proposed,so that the terminal equipment can also wirelessly charge the equipment while offloading tasks.Therefore,the optimization problem of the network is designed to provide the offloading decision for the terminal device task according to the condition of wireless channel in real time.Considering the complexity of combinatorial optimization,a deep reinforcement learning based offloading algorithm was proposed,which learned binary offloading decision from experience through deep neural network,thus eliminating the need to solve combinatorial optimization problem and greatly reducing the computational complexity.2)Inspired by the high flexibility and controllability of UAVs,a multi-UAV-assisted "cloud-edge integration" network architecture is proposed to offload computing-intensive tasks in terminal devices,in which UAVs can provide near-user edge computing.Based on this architecture,the computational offloading problem will be a mixed integer nonlinear programming problem,and it is usually difficult to obtain the optimal solution.Therefore,an efficient computational offloading algorithm based on importance sampling deep reinforcement learning is proposed to obtain the optimal computational offloading strategy.The algorithm sets the priority of each sample according to the target value and the actual value,so that speed up the fitting speed of the neural network.3)A mobile edge computing visualization platform based on Unity3 D and 3Dmax is proposed,in which 3Dmax is responsible for drawing the model,and Unity3 D is responsible for writing the logic functions corresponding to each model.During the operation of the platform,according to the terminal device selected by the user,the task information of the device and the offloading decision of each device task will be presented on the display panel.In addition,after each task is processed,the processing results are also saved in the database.