| As communication technology enters the 5G era,Internet of Things(Io T)applications that require real-time computation such as navigation,autonomous driving,virtual reality(VR)and augmented reality(AR)are emerging,which puts forward strict latency requirements for future computation networks.Mobile edge computing(MEC)is considered to be one of the key technologies in current and future communication and computation networks because of its ability to reduce computation latency by deploying computation nodes on the edge of networks.In MEC systems,limited and multi-dimensional resources of edge nodes and dynamic changing environment bring challenges to the system performance.In view of the above challenges,this thesis combines edge caching and other technologies to design and optimize the MEC system,and carried out in-depth researches on the resource allocation problem of the cache-enabled MEC system.Specific research contents are shown as follows:Firstly,the resource allocation problem of the MEC network is studied.In view of the challenges brought by dynamic task characteristics,fast-changing wireless environment and limited resources in edge computing system,this thesis introduces an unmanned aerial vehicle(UAV)into the edge computing system as a mobile edge server,in order to break through the limitations of traditional static edge servers.Moreover,the influence of UAV resources on system performance is analyzed,and the resource scheduling problem of the system is established by considering UAV hotspot selection and task unloading strategy.In further,the deep reinforcement learning algorithm is used to optimize the resource allocation problem through continuous interaction with the environment,which significantly improves the utility of the edge computing system.The simulation results show that the proposed scheme can effectively explore the appropriate hotspot selection and offloading strategy,and significantly improve the utility of the edge computing system.Secondly,based on the research of the MEC network,this thesis further combines with the edge caching technology to study the caching and computing resource allocation in the cacheenabled MEC network.Aiming at the problem of the system overhead caused by repeated requests and computations of tasks in the edge computing system,this thesis introduces a caching mechanism in edge computing system,and designs an effective cache replacement algorithm to reduce the system latency and energy consumption caused by repeated computation.Moreover,the cache algorithm is combined to optimize the system computation offloading,which achieves the goal of reducing the system latency and energy consumption.Simulation results show that the proposed method significantly improves the performance of edge computing system and achieves lower system latency and energy consumption compared with traditional methods.Finally,the profit maximization problem is studied from the perspective of resource provider for the cache-enabled MEC network.In view of the specific vehicular edge computing scenario,this thesis jointly considers the impact of the caching and computing resources,and models the profit maximization problem of the edge server.Moreover,cross entropy(CE)optimization algorithm is used to optimize the problem,and near-optimal caching and computing offloading strategies are obtained with low complexity,so as to achieve the goal of maximizing the profit of the edge server.Simulation results show that the proposed method is lower than the traditional Branch and Bound(Bn B)method in computational complexity and achieves the goal of maximizing the profit of edge server. |
