Energy-Efficiency Optimization For UAV-Enabled Mobile Edge Computing Systems

Mobile edge computing(MEC)is now an emerging paradigm that leverages cloud computing services to extend mobile base stations to the edge of the network,effectively avoiding the high latency and load generated by the transmission of large amounts of data in the cloud,while meeting the growing resource demands of internet of things(Io T)devices.In addition,unmanned aerial vehicle(UAV)is studied as a flying edge node and is used to improve the connectivity of terrestrial Io T devices.UAVs greatly facilitate the data collection and processing of Io T devices in MEC system due to their high environmental flexibility.Though the UAV-enabled MEC system is in close proximity to the users,it is still limited by energy resources.The energy consumption plays a prominent role in the UAV-enabled MEC system.Moreover,in order to extend the lifetime of UAV,the deployed energy resources must be fully utilized.To this end,this thesis focuses on two major aspects in the UAV-enabled MEC system: the relationship between the power consumption of the UAV and the system performance,and the issue of task offloading and resource allocation over collaborative multi-UAVs.The specific work and innovation points are as follows:(1)In a single-UAV MEC system,the UAV is viewed as an edge server,and this thesis studies the tradeoff between the power consumption of the UAV and the system stability.Specifically,this thesis firstly proposes a nearly complete system model,which includes a task queue model,an energy consumption model,and performance metrics(i.e.,the throughput and stability of the system,and the stability of the UAV battery power).Subsequently,this thesis combines the metrics of energy consumption and system performance,and then presents an optimization objective aimed at maximizing the time-average gain of the system under two serious conditions: system queue and UAV battery queue stability.Using Lyapunov optimization techniques,the original problem is decomposed into independent subproblems.This thesis separately resolves these subproblems by leveraging an online control algorithm.Then,this thesis provides mathematical analysis to guarantee the solution quality of our proposed algorithm,where the system revenue is theoretical maximized.Finally,extensive simulations are conducted to estimate our algorithm,and the experimental results well corroborate our theoretical analysis.(2)In a collaborative multi-UAV MEC system,this thesis studies the issues of task offloading and resource allocation.This thesis targets toward the tradeoff among the energy consumption generated by Io T devices,the energy consumed by UAVs,and the system performance.Merging the metrics of power consumption and system performance into a joint optimization objective,this thesis designs an online control algorithm to maximize the time-average gain while satisfying the system stability constraint.By adopting Lyapunov optimization technologies,this thesis formulates the original problem as independent subproblems.To take advantage of the separable reformulation,these subproblems can be solved with low computational cost.Furthermore,both rigorous mathematical analysis and extensive experimental results show the correctness and efficiency of our proposed algorithm.