Research On Joint Optimization Of Task Offloading In Multi-UAV-Assisted Mobile Edge Computing Network

In recent years,with the rapid development of the Internet and the Internet of Things technology,the number of terminal users(TUs)and intelligent devices have increased significantly,and the resulting computation-intensive and time-delay sensitive applications have generated huge computing demands,all of which accelerate the standardization process of Mobile Edge Computing(MEC).TUs always have real-time mobility and distribution uncertainty,and in the case of complex terrain or building obstacles,the function of traditional MEC technology is limited.Due to the high flexibility and adaptability of Unmanned Aerial Vehicle(UAV),UAV can realize self-organizing and adaptive networking technology to meet the needs of complex environments.UAV-assisted MEC networks have attracted the attention of industry and academia.Current researches on UAV-assisted MEC network are mostly limited to single UAV.Different from the limited computing power of single-UAV-assisted MEC network,multi-UAV collaborative networking can greatly improve the system performance,so as to cope with a more complex network environment.Therefore,how to properly coordinate multi-UAV for task offloading and resource allocation under the realistic scenario of TUs movement is an important problem that needs to be solved urgently in the current multi-UAVassisted MEC network.The main work of this thesis is as follows:(1)The task offloading delay optimization problem in multi-UAV-assisted MEC system is studied.Firstly,a physical model of static UAV as a fixed base station and dynamic UAV auxiliary relay is proposed to complete the computing tasks of real-time mobile TUs.Secondly,an optimization problem aiming at the total system delay was constructed to jointly optimize the TUs,static UAV scheduling,task assignment and dynamic UAV trajectory.Then,by formulating the optimization problem as Markov decision process one,the Deep Deterministic Policy Gradient(DDPG)algorithm is applied to obtain the optimal strategy.Finally,the simulation results verify the effectiveness of the proposed scheme,which can not only effectively reduce the total system delay,but also achieve fast convergence and better performance.(2)The energy consumption optimization of task offloading in multi-UAV-assisted MEC system is studied.Firstly,a three-layer multi-UAV-assisted MEC network is proposed,in which the High-rise UAV serves as a fixed charging point to receive and store solar energy,and the Low-altitude UAV(La UAV)cruises to provide MEC services for the real-time mobile TUs.Secondly,an optimization problem of total energy consumption based on load fairness was constructed to jointly optimize the3 D trajectory and task allocation of La UAVs.Then,the optimization problem is formulated as a Markov game process,and an improved Multi-Agent Deep Deterministic Policy Gradient(MADDPG)algorithm is applied to obtain the joint optimal strategy.Finally,the simulation results show that the proposed case can not only minimize the total energy consumption of the system,but also effectively optimize the 3D trajectory of La UAVs and ensure the fair load of each La UAVs.