Model Design And Simulation Of Collaborative Placements And Routing In VoD System

With the increasing and various need of study, work and entertainment, video service related applications has begun to boom and dominate the explosive development of network technology evolution and change. According to Cisco forecast report, the traffic of mobile video will contribute to70%of the whole mobile Internet traffic by2018. The system of video on demand, among all the video service, appeals to the users with the advantages of easy use and convenient operations. However, with the increasing number and the large size of videos, the library of videos is more and more difficult to be stored in a single server for the lack of space. How to effectively put forward a proper video placement is becoming a hot spot of research.To cope with the challenge of expanding video library, the solution of collaborative placement is introduced. The basic idea is to make full use of the high-speed link in the inner network of VoD system to store the video library collaboratively. As a result, each server stores only a subset of the video library. After the collaborative placement has been accomplished, when a video request arrives, the server would send the video data directly to the users if the video is stored on it; otherwise, the video request would be completed by collaborative routing from the origin servers, which store the video. Based on the above point of view, the collaborative placement and the collaborative routing were studied. The main contents are as follows:1.To begin with, we introduce the characteristics of VoD system, such as the distribution of video visiting frequency, user visiting pattern, the workload in the system and so on. Besides, we introduce the related literatures about video collaborative placement, and point out the shortcomings existing in them. Then the research ideas has been put forward, the model consists of two aspects:collaborative placement and collaborative routing.2.As with the collaborative placement, an algorithm has been proposed to solve the maximum local hits problem based on video segments in the collaborative placement model. With joint study of network coding and video collaborative placement, the algorithm constructs a special network flow graph and converts the original problem as the minimum cost flow problem over the graph. Thus, the optimal solution could be obtained.3.As with the collaborative routing, a greedy algorithm has been put forward to minimize the maximum link utility in the collaborative model after the video segments has already been placed on the servers. The video requests would be processed in the time rounds. At the start of a time round, the video requests are collected, then dealt with at the end of the time round to choose the source server with minimum link utility and begin the data transmission.4.Extensive simulations are carried out to evaluate the performance of the proposed algorithms. The results in collaborative placement experiments suggest that the proposed algorithm outperforms the other existing solutions. The influence of the segment size in the algorithm and whether the proposed algorithm works in different video frequency distributions have been analyzed too. The results in collaborative routing experiments show that the proposed algorithm outperforms than the random server selection algorithm and nearest server selection algorithm, it also works well in different network topologies.