Resource Allocation In UAV-aided Edge Computing

With the advent of 5G era and the vigorous development of the Internet of things industry,a large number of intelligent devices with network access demand have been generated.As a result,the amount of data in the Internet has exploded.There are more and more network services with more stringent requirements for network real-time,reliability,computing capacity and security.Edge computing emerges as the times require.Edge computing brings computing resources to the edge of the network closer to the user terminal,and processes the user business nearby,which greatly reduces the network communication delay and improves the network security.In addition,due to the wide distribution of various intelligent devices,the traditional fixed base station is subject to geographical factors sometimes it can not provide good signal coverage,which restricts the development of edge computing.Unmanned aerial vehicles(UAVs),which are costeffective and easy to deploy,are used to carry computing platforms,which give computing platforms the high mobility and overcome communication barriers caused by geographic factors.The edge network generally does not have the enormous computing resources as the cloud data center,especially the platform carried by UAV.Unlike the base stations,UAV can’t easily be charged.Meanwhile,the movement mode of UAV is also complex and diverse.Therefore,many factors need to be considered in resource allocation,and how to allocate resources is very complex.So this thesis makes an in-depth study on the problem of edge computing resource allocation with UAV-aided.(1)For the scenario where a single UAV serves multiple ground terminals,we consider the UAV doing periodic circular cruise,and design the cruise radius and altitude of the UAV to serve a larger range of ground terminals.The task load offloading ratio,ground terminal and UAV computing resources are also optimized.The system’s average delay optimization problem is established under the constraints of power consumption and computing resources,and the problem is solved iteratively by a combination of problem relaxation and branch and bound algorithm.The convergence and global optimality of the solution algorithm are also proved.(2)For the scenario where multiple UAVs serve multiple ground terminals,we consider allowing the UAV to hover at certain locations and design their hovering coordinates to serve ground terminals,and the task load offloading ratio are also optimized.The optimization problem of system average time delay under the constraints of energy consumption is established.The problem is solved iteratively by a combination of variable substitution,problem relaxation and branch and bound algorithm,and the convergence and global optimality of the algorithm are analyzed.(3)This thesis verifies the algorithm of the above two scenarios through numerical simulation,and analyzes the relationship between the change of input parameters and the result.It shows that when the trajectory of the UAV meets a certain relationship with the geographical distribution of the ground terminal,the average processing delay of the system is lower.It also shows that the more unmanned aerial vehicles and computing resources,the more tasks the ground terminal can offload,the more improvement of the system delay.