Efficient Collaborative Caching In Cloud Radio Access Network

In recent years,the data transmission in wireless communication network is growing exponentially,which brings severe challenges to wireless communication system.How to deal with the rapid growth of data traffic so as to optimize the content transmission and reduce the service delay has become a research hotspot in academia and industry.As one of the key technologies of 5G network architecture,Cloud Radio Access Network(C-RAN)is a promising architecture with high capacity,low energy consumption and low cost.Active caching of popular content in C-RAN can further improve the quality of service(Qo S)and quality of user experience(Qo E).As a kind of terminal direct communication technology,D2 D can realize the direct data transmission between two adjacent devices,and avoid the transfer of data through the base station,which is one of the key technologies of 5G mobile communication.When D2 D is added to the cache content distribution,users can obtain the cache content from the adjacent users,which effectively expands the coverage of the base station.Based on the C-RAN architecture and D2 D communication technology,this thesis designs efficient edge caching and distribution mechanism.The main work of this thesis can be summarized as follows:1)Based on the typical C-RAN architecture,a collaborative caching mechanism is proposed.In this mechanism,popular contents are cached on both BBU and RRHs.The vertical collaboration between BBU and RRHs and the horizontal collaboration between RRHs are used to deploy content cache.In order to minimize the average transmission delay,an optimization model is established and solved by two parts.Firstly,based on the time series of content visits,the autoregressive model is used to predict the content visits and calculate the content popularity distribution.Secondly,based on the local content popularity distribution characteristics,an improved genetic algorithm is proposed to solve the cache deployment problem.Through the simulation experiment,the prediction effect of the AR model is first evaluated.Then,compared with the traditional genetic algorithm(GA),the most popular cache(MPC),hybrid cache(HC)and random cache(RC)schemes,the average transmission delay of the proposed scheme is reduced by 33.3%,21.3%,37.6% and 49.7% respectively without considering the prediction accuracy of content popularity.Finally,the predictive value of content popularity is applied to the cache deployment to verify the feasibility of the step-by-step solution mechanism.2)On the premise that the base station has deployed the cache contents,this thesis then proposes a cache content distribution mechanism in the mobile scenario.Due to the mobility of users,the success rate and delay of getting cache content from the base station cannot be guaranteed.Therefore,this thesis proposes a D2 D assisted fragmentation caching and distribution mechanism.The base station and mobile users within the base station's communication range are all considered as possible transmission sources.This thesis firstly classifies the possible transmission source nodes based on the relative distance and relative speed,and adopts the idea of risk aversion to predict the file slice transmission rate of various transmission sources over time.Thus,the file slice transmission time axis of each transmission source is obtained.Based on the predicted time axis,according to certain criteria,the transmission source node is reasonably arranged to cache or distribute the content at specified time.Simulation results show that the proposed mechanism can improve the success rate of transmission as well as reduce the transmission delay.Compared with the random node transmission scheme,the transmission success rate of the proposed scheme is improved by 13.7% ? 58.5%,and the average transmission delay is reduced by 21.9% ? 53.4%.