The Research Of Caching In Fog Radio Access Network

The core of fog radio access network(F-RAN)is to effectively alleviate forward traffic by allocating edge caching resources and computing resources reasonably,which results in achieving the goal of reducing latency and improving energy efficiency.In F-RAN,user mainly offloads data traffic from small cell network and the traffic is mainly composed of video.However,for edge caching to deal with such traffic offloading,the current researches underperform on latency and energy efficiency improvement.Therefore,this thesis studies the file popularity of videos and the caching strategy of video files in small cell network with limited caching capacity,further proposes an optimized prediction algorithm of file popularity,as well as latency caching algorithm and energy efficiency caching algorithm.The prediction accuracy of file popularity directly affects the improvement of network performance by caching strategy.Existing methods for predicting the file popularity of videos mainly depend on the number of request.They update the file popularity of videos only by the number of request without analyzing user's favor and file attribute,which results in inaccurate prediction accuracy.This thesis presents a prediction algorithm of file popularity based on file attribute awareness,which extracts features from video tags and tags file with features.Further,it establishes the relationship between user's favor and file features through feature weights.Hence it can learn the files of user request online and predicts the file popularity in real time and accurately in each slot with the consideration of temporal and spatial factors.This thesis studies the delivery process of file in two small cell networks,focusing on multi-point cooperative transmission in non-cluster small cell network,joint transmission(JT)mode and parallel transmission(PT)mode in cluster small cell network.For non-cluster small cell network,this thesis presents an optimized latency caching algorithm.Based on the latency model of non-cluster small cell network,the problem of latency caching is transformed into a location problem of file caching.The problem is transformed from non-convex to convex by variable relaxation,and the final caching strategy is obtained by linear programming and variable recovery.In this thesis,an optimized energy efficiency caching algorithm is presented for cluster small cell network.Firstly,the model of energy efficiency caching problem is established,and the successful delivery rate of JT mode and PT mode is analyzed.Then,the problem is decomposed into two sub-problems: maximizing cache hit rate under transmission consumption restriction and whether convert file into the most popular file under energy consumption restriction,which results in realizing the energy efficiency caching strategy with low algorithm complexity.The simulation results show that based on file attribute awareness,the proposed latency caching algorithm and the energy efficiency caching algorithm can accurately predict the file popularity,and can significantly reduce the user latency in non-cluster small cell network and improve the system energy efficiency in cluster small cell network,respectively.