Research On Cache Based Cloud Radio Access Network

With the development of hardware and the tremendous amount of mobile traffic, traditional cellular network architecture is not adequate to meet the increasing user demands. Fog computing is used in the heterogeneous cloud radio access network to meet the demands. By centralizing the processing unit and combining with the real-time cloud computing, the efficiency is greatly improved and the cost for deployment is decreased. However,plenty of network traffic is derived from a lot of popular contents that have been requested multiple times. The traffic load caused by these contents can be alleviated by deploying cache at the edge of the network. In this thesis, the cloud radio access network architectures and the deployment of cache in the current networks is investigated. The main contribution of this thesis is described as follows:Firstly, the evolution of the cloud radio access network architecture and the cache techniques is studied. Then the deployment of cache in current networks is described, and the common metrics of network performance with cache is discussed.Secondly, caching at the cloud base station is studied. By introducing the cache enabled access point, the chunk based granularity for content caching is applied in the proposed architecture. The aim is to find the distribution of the contents among C-APs to optimize the average hit rate of user devices. A mathematic model for caching strategy is proposed, and the problem is formulated as a constrained convex problem,which can be solved by a binary search algorithm with low complexity. The proposed strategy has been compared with multiple cache strategies, and the simulation results validate the effectiveness of the performance in term of the average hit rate.Finally, a user interest based cooperative caching strategy is proposed.By introducing the concept of interest matrix and the similarity of interest matrix, the workflow of the user equipment cooperation is described, and the strategy's theoretic average hit rate has been analyzed, which is validated by the simulation results. Then the influence of user equipment density, cache size and the threshold of the similarity of interest matrix to the performance of the strategy is explored.