Research On Energy Optimization And Task Allocation In UAV-Aided Edge Computing

Edge computing(EC)is a promising technology which deploys the edge nodes with high computation capability near the ground clients(GCs)to reduce the transmission delay.Currently,most works in EC mainly focus on deploying stationary edge nodes(such as road side units,femotocells)to assist the BS to complete GCs’ tasks.However,when it comes to emergency situations(such as earthquake,or the fire)or large numbers of GCs change their locations,it is hard for stationary edge nodes to complete GCs’ tasks in time.As a low-cost aircraft,unmanned aerial vehicles(UAVs)has advantages of high mobility and flexibility.Placing edge computing units on UAVs can provide the communication and computation assistance for GCs that are lack of computation capability,which realizes the cooperation among the base station,UAV and GCs.There are mainly two challenges in UAV-aided edge computing.First,UAVs’ flight time is fundamentally limited by their maximum battery capacity.How to design an algorithm that jointly optimizes UAVs’ flight trajectories and flight speeds to reduce their energy consumption is important.Second,the computation resources for GCs are limited.How to design a taskallocation algorithm that appropriately schedules the resources to the BS,UAVs and GCs to improve efficiency of task processing is another challenge in this work.To solve above challenges,we conduct the following research works:First,we formulate an energy-efficient speed scheduling problem for situation that GCs are deployed in linear.The core idea of this problem is to adjust the UAV’s flight speed to minimize its energy consumption within task’s dealine.To this end,an offline optimal algorithm based on convex optimization is proposed to find the boundary flight speed for the UAV.We prove that the proposed algorithm optimally solves the speed scheduling problem in theory.Then,an energy-efficient flight planning problem is studied for situation that GCs are deployed in a 2-D platform.The core idea of this problem is to minimize energy consumption of the UAV by jointly optimizing its trajectory and flight speed.To solve this problem,we design a flight planning algorithm consisting of UAV’s initial trajectory,speed scheduling and trajectory adjustment to save UAV’s energy under the constraint that all GCs completes all their tasks in time.Simulation results demonstrate that our proposed algorithm saves energy both in trajectory designing and speed scheduling.Afterwords,a tripartite cooperative task-scheduling problem including the base station(BS),UAVs and GCs are designed to maximize the GCs’ task completion amount.The problem focuses on jointly optimizing UAVs’ trajectories and scheduling GCs’ tasks under the limitations of UAVs’ maximum battery storage.We propose a trajectory design algorithm based on prescheduling technology and a task-schedulign algorithm according to primal-dual research.The competitive ratio of the task scheduling algorithm is proved to be 1.58 in theory.Finally,based on above three theoretical solutions,a UAV-aided edge computing system aiming at minimizing UAV’s flight energy and maximizing GCs’ task completion amount is designed.The sytem realizes the above three theoretical solutions in practical environment,the usability and effectiveness of this system are also verified.Compared with the existing works of UAV-aided edge computing,the algorithms studied in this thesis perform better in terms of speed scheduling,flight planning and task scheduling,which save the energy consumption,prolong the flight time of UAVs,and improve GCs’ task completion amount.With the wide application of UAV-aided edge computing,the proposed algorithms can be further utilized in emergency communication,industrial Internet,smart cities and many other fields.