Design And Realization Of Layered Video Multicast System

With the development of the multimedia technology and wide band networks, video multicast is widely applied in many fields such as video conferencing, distance learning and living video broadcasting. Rather than send a copy of each packet to each user individually, the multicast sending an only packet to a group of users. While in the heterogeneous networks, the receivers of the video multicast can receive video data at different rate. The service with uniform rate will leads to congestion in some links while some bandwidth is unused in other links. And because most of the popular video encoding standards adopts the inter-coded frame, the loss of a key frame will cause the failing decoding of several succedent frames. So it is significative to search on a layered video multicast system in which the raw video data is encoded into several layers and delivered and users can receive video data at different rate commensurate with their respective bandwidth. The thesis summarizes the theory of video encoding/decoding and multicast and analyses the limitation in video multicast over IP networks. And based on the research on the scalable video coding, the application of real-time transfers and the technology of QoS, the thesis presents a receiver-driven layered video multicast scheme and realizes the archetypal model. In the model the video data is encoded into several layers in the video server. The layer with most important features, called basic layer, can be decoded independently while additional layers, called enhancement layers, can progressively refine the reconstructed video quality. The hierarchical video streams are delivered with layer mark that corresponds to the service level. The users adjust their reception rate by choosing appropriate service level. The higher the level their apply, the more layers their will receive. RSVP is used to aggregate the apply level from different links and distribute the stream to the links according to the service level at intermediate nodes. And when network load changes, user adjusts the applying level according to the QoS of the network to obtain preferable service. The result of the exam shows that the receiver-driven layered video multicast can guarantee the transmission of the most important features and provide multi-rate service and goodish QoS.