The Study Of Active Layered Multicast With Resource Reservation

The video stream delivered over network is becoming more and more, it becomes the basic requirement for network to support the video stream delivery, as one of transmission methods, multicast is used widely. Video stream have some features such as massive data, high bandwidth, delay and jitter sensitive, but the network is highly heterogeneous and dynamic. All this make congestion control mechanism becomes a key policy for video stream multicast. As congestion control mechanism, layered multicast and resource reservation could be used to improve the quality of service of the network. In active network, new protocol can be deployed easily and the active router has complicated functions, a new function could be extended rapidly. In this paper, a scheme about active layered multicast based on resource reservation is discussed and the simulation result is given.The scheme can work in active network, the video stream is coded into different layers, the required resource is reserved. The scheme has a few features.The video data is coded into different layers at the source and only one multicast group is used.The receiver initialed the reservation request. The request with the same attribution can be merged in routers.In this paper, the basic theory of multicast, receiver-driven layered multicast (RLM), active network and resource reservation is introduced. After that, some other scheme about layered multicast and active layered multicast is analyzed clearly. The mechanism about multicast simulation inside network simulator NS2 is given. Some source codes and data structures is analyzed. An extension of NS2 for active multicast network is discussed and implementation is given. Based on this extension, many tests have been done. The test results are compared with the results of RLM and the reasons is given.At the end of this paper, the disadvantages and the possible improvement of the scheme are discussed.It is necessary to point out that the scheme discussed in this paper is only for video stream, in which some redundant information could be dropped.