Cooperative Caching Strategy Of Multiple Base Stations Based On User Request Content Prediction

With the rapid growth of mobile data traffic,operators are deploying small base station(SBS)intensively to meet the demands of peak network traffic,so the number of SBSs and users under one remote server is increasing.The base station obtains the user's request content from the remote server via the backhaul link.However,the limited resources of backhaul link can't meet the needs of all users in time,limiting the improvement of network performance and user's quality of experience(Qo E).Actually there are many repeated requests for the same content in network,the repeated transmission of the same content not only wastes the limited bandwidth resources of the base station,but also user's delay of obtaining the content is large.The edge caching solves the shortage of backhaul link and the improvement of user's Qo E by reducing the distance between users and contents.By caching the content on the base station,when user's requests hit the cache,user can obtain the content from the base station directly.However,the cache space of base station is much smaller than the total content of the remote server,and mobile user associates different base stations at different times.Therefore,it's worth studying the question that optimizes the selection of base station cache content to reduce the user's delay of obtaining content.Due to the limited cache space of SBS and the change of user associated base stations at different time,this thesis proposes a multi-SBS cooperative caching strategy based on user's request content prediction.First of all,in this thesis,the cache placement strategy is mathematically modeled with the goal of minimizing the user's average delay of obtaining content.Then,macro base station(MBS)divides the fixed cooperative set according to the location of SBS,under the premise that the user's movement trajectory is known,the global content popularity and user's request characteristics mined from historical request data are used to predict the content that user is most likely to request.When the associated SBS and its cooperative SBS are not cached and the associated SBS has available cache space,it is cached in the associated SBS.If the associated SBS hasn't available cache space,perform the cache replacement method.According to the content cache value decide which content to replace in the cache area.The static division of the cooperative set only considers the location information of SBS,and during the entire period of user movement the division of the cooperative set is same.However different user may request the same content at the same time slot,so static partition is not the best solution to reduce content repeated caching.This thesis proposes a method to divide the cooperative set dynamically according to all users' s request content and associated SBS.In the current time slot it predicts the user's next time slot request contents and associated SBS.According to the criteria of dividing cooperative set,divide as many cooperative sets as possible from user associated SBS that requests the same content.This division reduces the numbers of caching content and takes less cache space to cache the same type content.Finally,the scheme in this thesis is compared with the existing scheme.The simulation shows that the proposed scheme can reduce the user's average delay of obtaining content,improve the ratio of cache content types and used cache space,reduce the times of caching content at different SBS and lower the cost of caching.