Research On Intelligent Caching Mechanism Of Air-ground Cooperative Networks

UAVs-assisted caching of air-ground cooperative networks is a key component of mobile edge computing(MEC).It utilizes the caching resources of UAVs to cache contents at the off-peak time,and distribute contents to users at the peak time.The storage and capacity of the battery mounted on UAVs are limited,the placement strategy of caching content will influence the content hit ratio(CHR)and content delivery delay.Thus,the problem of selection and placement of caching contents is studied with the consideration of improving the energy efficiency(EE)of UAVs in this thesis.Firstly,we propose an intelligent caching scheme for the base station(BS)with the assistance of UAVs.In this scheme,a content popularity prediction algorithm is proposed to predict the popular degree of content among users.Then,the deep reinforcement learning(DRL)algorithm is utilized to optimize the placement strategy for caching contents.Based on the user requests,the hidden Markov model(HMM)is adopted to effectively schedule UAVs in the content delivery process.Secondly,to alleviate the challenge of the limited caching capacity of UAVs,a UAV-assisted caching and tasks offloading scheme is proposed in device-to-device(D2D)communication networks.In this scheme,the local storage resources of terrestrial users are used to cache the content.In addition,based on the analysis of the content popularity,a user activity model is proposed to describe the content popularity more accurately.Moreover,to maximize the data transmission rate and minimize the task processing delay during task offloading,the multi-agent reinforcement learning algorithm is applied to optimize the users’ computational task offloading strategy and the deployment of UAVs.The aforementioned two caching schemes in UAV-assisted networks are evaluated by simulation.The results illustrate that the proposed intelligent caching schemes outperform the existing solutions in improving the CHR and reducing content delivery delay.Meanwhile,the EE of UAVs can be enhanced by rational deployment and scheduling of UAVs.