Research Of Mobility Aware Edge Caching Strategy In Wireless Networks

With the increasing demand for wireless broadband services,wireless networks are facing increasingly pressure on data traffic.How to perform traffic offloading on cellular network systems under the premise of ensuring user experience has become a research focus of the industry.Introducing content caching mechanism in cellular and D2D(Device-to-Device)hybrid network can effectively increase network capacity,reduce backhaul link burden and content transmission redundancy.Due to the contradiction between user selfishness and D2D communication sharing,under the mobility scenario,caching policy setting has the problems that file popularity and user mobility are imperceptible and inability of base station caching.Also,the storage optimization in cellular networks has the problem of limited caching performance due to inadequate matching between base station caches and user requests.To solve above issues,this thesis studies how to perceive the user mobility and file preference information and proposes an intelligent caching strategy.This strategy can solve the problems of how to set incentive and recommendation scheme and how users and base stations' caching in the D2D heterogeneous network under different mobility scenarios.The main tasks and contributions of this thesis are as follows:1.Research on caching strategy based on user mobility and incentive scheme in D2D network.A user caching strategy based on incentive scheme is proposed according to the mobility of successful communication time distribution model between users.This caching strategy solves the problems of maximizing users' utility and minimizing base stations' cost at the same time.The user caching strategy and the optimal unit reward provided by the system can be obtained by the gradient projection algorithm.The simulation result shows that the optimal unit reward increases as mobility speed decreases.The proposed caching strategy can reduce base stations' serving cost up to 32.8%and 13.1%compared with greedy and incentive caching strategies respectively in four different user mobility scenarios.The advantages of reducing base stations'serving cost become more obvious as user mobility speed increases.2.Research on intelligent caching strategy based on recommendation scheme in cellular network.An intelligent caching strategy of base stations based on recommendation scheme is proposed.The Q-learning algorithm is used to learn user mobility and the regularity of users' application for files,and a base station caching strategy aiming at minimizing long-term base station cost is obtained.This part answers questions about how a base station should develop a caching strategy and how to set up a recommendation scheme under different file preferences in a cellular network.The simulation results show that the lower the user mobility rate,the weaker the optimal recommendation strength is.The mechanism can greatly increase the hit probability of both D2D network and small base stations in different mobility scenarios.The proposed caching strategy can greatly reduce the core network cost up to 22.5%and 18.6%compared with greedy and recommendation caching strategies respectively in four different user mobility scenarios.3.Research on intelligent caching strategy of temporal and spatial based mobility model in cellular network.The shortcoming of the above research is that the accuracy of the user moving trajectory modeling is insufficient,so that the resources of base stations on the moving traj ectory cannot be fully utilized.Therefore,we use temporal and spatial based mobility model to model user mobility,and to maximize the transmission success rate as the optimization goal,we learn user mobility and their preferences for files.The Knapsack algorithm is used to compress the action space,and the Q-leaning with Knapsack algorithm is applied to give the caching deployment guidance of the base stations in different mobility scenarios.Simulation results show that the temporal and spatial based mobility model has greater system reward advantages than the single-dimensional mobility model in the proposed caching strategy.Also,the lower the user mobility rate,the weaker the optimal recommendation strength is.The proposed caching strategy can highly improve system benefits up to 29.8%and 15.2%compared with greedy and recommendation caching strategies respectively in four different user mobility scenarios.