Research On Video Prefetching In Mobile Social Network Based On D2D Communication

In recent years,with the rapid development of mobile communication,mobile social applications are developing rapidly.Sharing of multimedia data(video,audio,text,etc.)is one of the important applications in mobile social networks.Mobile users can use portable wireless communication devices to create and share multimedia data freely and conveniently.The increasing number of users have been contributing to the rapid growth of multimedia data,resulting in a huge amount of traffic in cellular network.Among the multimedia data,video data requires the most network transmission resources.Deviceto-device(D2D)communication is an important method to alleviate the burden of cellular networks.In order to use D2 D communication to transmit the videos in online social networks,two problems need to be solved: one is to predict which users will watch the videos in social network;the other is how to transfer the videos from the source users to the target users through multi-hop D2 D communication.This thesis focuses on the research of video prefetching prediction and D2 D transmission of video data.The main research contents are as follows:(1)Research on video prefetching prediction based on social relationship and video content.The purpose of video prefetching prediction is to predict whether the video will be viewed by the users,and it is the basis for video prefetching using D2 D transmission.Existing video recommendation algorithms mostly focus on the TOP-N problem,which recommends multiple videos to the users.However,limited literature studies the problem of whether the video will be watched by the users.We propose a Social-and Content-aware Video content delivery Prediction method(SCVP).We design five metrics to evaluate active degree of users,social tier between users,similarity between videos,similarity between user interest and video content and video popularity.Combined prediction is used to integrate the impact of the five factors on the prediction.Experimental results show that the proposed method SCVP can effectively improve the accuracy of video prefetching prediction.(2)Research on D2 D transmission of videos based on user's encountering history and social features.Due to the mobility of social users,it is difficult to find a D2 D path that can stably transmit the video to the target user.That is,the data may not be successfully transmitted to the target user.Therefore,efficient routing algorithms are essential to improve the performance of video data transmission via D2 D.It is an important issue how to select appropriate mobile social users as relay users using D2 D to transmit the videos.Existing routing algorithms mainly design routing algorithms based on the user's encountering history or the social media content.Limited literature captures the physical location proximity between users reflected by users' social features.We first introduce a relay node selection metric that integrates the user's social features and user's encounter history.We then propose a Routing based on Encountering history between users and Social features of users(RES).Experimental results show that algorithm RES can effectively improve the performance of data transmission success rate,average transmission delay,and average number of hops.