Online Hot Spot Video Cache Replacement Policy Based On Popularity Prediction In High Density Heterogeneous Networks

The dense heterogeneous network is a new network architecture proposed for the future mobile communication networks.Dense heterogeneous network can improve the performance of wireless networks by reducing the distance between base stations and users.As more and more users like to watch online short videos anytime and anywhere,dense heterogeneous networks will carry large-scale video data traffic.In order to alleviate the pressure on the base station,the application of caching technology in dense heterogeneous network is an effective solution.The popularity of online video is an important attribute of dense heterogeneous network caching.Firstly,we obtain the online video views by crawlling the Bilibili website,and analyze the changes of the online video views and the proportion of online hot spot video.Data analysis results show that the popularity of online videos is highly dynamic,and approximately 20% of online videos accounts for 80% of total views.Then,by dividing the data set of online video views into the data set of the time series of length 11,we construct a k-nearest neighbor regression model for popularity prediction.We also analyze the scheme of popularity prediction by the Affinity Propagation Clustering algorithm and make a comparative experiment.The experimental results show that the k-nearest neighbor regression model has a small mean square error and good stability,showing good prediction performance.Finally,we propose an online hot spot video cache replacement policy(OVCRP)based on popularity prediction in high density heterogeneous networks with the goal of minimizing the overall transmission delay.For the mathematical model of the policy,the graph algorithm is used to solve the problem step by step,including neighborhood user aggregation,redistribution of facilities,getting a half-integer solution and modifying the half-integer solution,and finally obtain the approximate integer solution of the mathematical model.Simulation results show that the OVCRP has a higher average cache hit ratio and lower average access latency than RANDOM cache scheme,the least recently used(LRU),and the least frequently used(LFU).