Research And Design Of Live Video System Based On Scalable Video Coding Over P2P Networks

With the increasing development of Internet, the video becomes the main carrier of networking information. According to statistics, more than40%of the network traffic is generated by streaming media. Although researchers have proposed a lot of relevant research achievements, there are still some crucial challenges in this field. First of all, the heterogeneity of both network environment and clients becomes a hinder to the popularization and development of network streaming media. The client access networks includes wired networks, Wi-Fi and3G networks which have different bandwidth; even the same network still suffers bandwidth fluctuation. The clients with different hardware require different video quality. Most of the current video coding standards can only provide fixed stream, which cannot dynamically adapt to the fluctuation of network and cannot meet the different requirements of clients. Besides, in traditional C/S mode networks, there are a large number of clients and a single server, which results in a heavy load on the server side. Hence, the networks with C/S mode have a poor scalability.To address the above problems, this paper designs and implements a live video system named NccLive over P2P networks based on scalable video coding (SVC). SVC encodes a single video source into multi-layered video flows with different video qualities, which can effectively relieve the problems caused by the heterogeneity of both networks and client hardware. But the existing open-source software JSVM cannot meet the real-time encoding requirements of network streaming media. Therefore, this paper optimizes the JSVM encoder from two aspects which include invoking the IPP functions and introducing the parallel processing. On the other hand, P2P technology can eliminate the unbalanced load problem in the C/S mode and provide good scalability for live video streaming system. P2P technology gets fully attention in the recent years. However, the existing scheduling algorithms and incentive mechanisms cannot be well applied in the video content distribution based on SVC To this end, this paper also proposes new data block scheduling strategy and incentive mechanism. Tests demonstrate that NccLive has good stability and can flexibly provide adaptive multi-level video services. The proposed system can be applied to the scenarios like live video and real-time monitoring, etc.