The Research Of Mobile Edge Computing Network Based Video Cache Algorithm

With the development of Internet and mobile communication technology,the number of mobile devices grows rapidly,and the data traffic of network increases exponentially,of which the video traffic occupies a large proportion.On the one hand,the emergence of new-type video services has led to a significant increase of the video traffic,on the other hand,the delay requirement of new-type services for video content access is becoming more and more stringent The rapid growth of video traffic and the requirement of low latency pose new challenges to current network architecture and various algorithms deployed on network nodes.In order to solve the problems brought by the growing business demand,many new technologies have been introduced into the fifth generation mobile communication network.As one of the key technologies of 5G,mobile edge computing technology brings more computing resources to edge network nodes,which can improve the intelligence of edge network and has a wide application prospect.Therefore,MEC has attracted great attention of researchers.Content pre-push is a common caching technology in current network.Its core idea is to use the low peak period of network traffic to push the content accessed by users to the edge node which is closer to the user in advance.In this way,the content accessed by users can be pushed directly from the edge node to the user in peak traffic,thus reducing the load of backhaul links and the network delay.The core algorithm of this technology is how to select the contents.Based on the traditional recommendation algorithm,this paper studies the relationship between video types and user ratings,which is one of the most easily accessible attributes of video requests data.Thus,a rating prediction algorithm based on video type and matrix factorization is proposed.The estimated rating obtained by matrix factorization is regarded as fine rating.The estimated rating based on video types is a rough rating.And the latter is used as an amendment to weigh the former.The amendment parameters is calculated from individual to individual.The simulation results show that the proposed algorithm can effectively improve the accuracy of rating prediction.With the development of hardware and network technology,user-end equipments have gradually evolved from the old mobile devices with weak computing power and low storage capacity to the smart devices with high computing resources and high storage capacity.This evolution also enables the user-end devices to play more roles in the network.The introduction of MEC technology makes the edge network closer to users more intelligent.And the opening of third-party interface also provides more tools for manufacturers to optimize their network applications.In this paper,a Joint Cache algorithm based on User Preferences(JCAUP)is proposed,which takes advantage of the computing power of MEC edge nodes and the high storage capacity of user-edn devices,combines the base station cache and device cache,and introduces the GSVD algorithm mentioned above.The solution uses the recommendation system to learn and discover the preference diversity of users,and then chooses to cache the content on the user-end device cache or the base station cache server.The simulation results show that the algorithm can effectively reduce the user access delay and enhance the user experience.