Abstract
In this thesis, a new approach for real-time video streaming on a peer-to-peer network is proposed. It provides the ability to control the quality of service of networks from end systems. This scheme makes use of the layer video coding techniques and extends them to the peers such that users can select the quality of video by themselves. The communication of the protocol is simply constructed by the request of receiving peers. Our system is adaptive to network bandwidth since users can choose different levels of video depending on the traffic conditions. We inserted two types of tags, the timing tag and layer tag, into the video packets for implementation. With the proposed management of the tags, the timing tag is used to reduce the display delay while the layer tag is used in a congested network. When serious congestion happens, the higher layers of video data will be sacrificed to get more bandwidth for lower layers of data. Therefore, our system is adaptive to the bandwidth. These functions provide the real-time ability and are adaptive to non-congestion and congestion conditions. Moreover, the server can provide real-time patching channels to stabilize the video quality of peers. We simulate this system by ns-2 with the proposed scheme, and show that the packet loss rate is reduced to around 10% with 1000 peers. The standard deviation of delay time is below 15ms. Besides, we also show the reduction of the maximum queue size and the adaptive ability of the video stream in this system.