User Behavior Analysis Method For Edge Caching In Wireless Fronthaul Network

With the rapid development of mobile Internet and the continuous prosperity of emerging network application services,the number of mobile network users is increasing dramatically.At the same time,users' demand for mobile Internet in large capacity,low delay and intelligence is also increasing.In order to meet the needs of users,the Fifth Generation Mobile Communication(5G)comes into being.The deployment of cache resources on the edge of 5G fronthaul network can effectively reduce the transmission delay of network contents and relieve the load pressure of network link,which is an effective way to improve the user experience of the network.However,most of the existing network edge caching schemes only consider the popularity of network content in a regional network or the whole network,but ignore the different components of network users and the differences between the network content requested by different network user components.To solve these problems,this paper proposes a hybrid edge caching scheme based on the analysis and prediction of users' behavior.The main research contributions are as follows:A Density Ratio Peak(DRP)user clustering algorithm based on the density ratio estimation method is proposed to solve the clustering and behavior analysis of users in 5G network.According to the preferences of network users,a Density Peak(DP)clustering algorithm is used to cluster network users into different clusters.However,when the density of user behavior data between the various clusters is too large,the DP algorithm cannot accurately identify all the clusters in the user data set,or the proportion of users classified into the wrong clusters greatly increases.To solve this problem,this paper improves the DP algorithm by using the density ratio estimation(DRE)method,and proposes a DRP user clustering algorithm based on the density ratio estimation method.According to the clustering results,it is necessary to predict the network content that users in each cluster will request.In order to improve the accuracy of the prediction results,this paper uses the auto regressive integrated moving average(ARIMA)model to predict the users' behavior data,and uses the empirical mode decomposition(EMD)algorithm to deal with the non-linear problems of the users' behavior data.And the simulation results show that the proposed algorithm is better than the traditional method.According to the results of user behavior analysis and prediction above,for single cell and multi cell application scenarios,this paper proposes active hybrid edge caching strategy and active hybrid cooperative edge caching strategy respectively.According to the clustering results of network users in each cell,the cache value of network contents applied by users of each cluster is predicted and calculated respectively,and the corresponding network edge cache strategies are developed accordingly,so as to comprehensively consider the specific needs of network users of each cluster.Compared with the traditional edge caching scheme,the proposed scheme in this paper can significantly reduce the network content transmission delay,reduce the cache redundancy and improve the cache hit rate.