IP多点组播视频的可靠扩展控制机制研究

The Research on Reliable and Scalable Control of Multicast Video over IP

针对 IP多点组播视频控制信息组播传输的可靠扩展问题 ,将组播扩展性方案超立方体 ( hypercube)拓扑思想与组播可靠性方案反馈重发局部化 ( localization)思想用于控制机制 ,提出了一种基于超立方体拓扑的可靠扩展控制机制 :将控制拓扑从 1维树型映射为 n维超立方体拓扑 ,运用超立方体结构几何特性将局部反馈重发恢复融入拓扑节点扩展中 ,实现了视频控制信息的有效可靠扩展 .理论分析与实际测试表明 :基于超立方体拓扑的控制机制有较好的可靠扩展性 .

Multicast video over IP, as an important application of network multimedia, is an active research field these years. One of the focuses is the reliable multicast of its control information in the scalable video nodes topology, for which multicast reliability and its scalability are not easy to be combined in the same control mechanism. This paper presents a hypercube-based multicast reliable & scalable solution, which is combined the hypercube character of scalable solution with the localization character of reliable solution to solve the problem.The solution can be described briefly in the following way: first embedding the 1-dimension tree topology of multicast video nodes into N-dimension hypercube topology by the Liebeherr solution, in which some proper modifications are made for the specific Multicast video application; then using some real-time dynamic management of the hypercube topology to guarantee the mechanism in the stable hypercube state and updating the nodes topology information by multicasting control information periodically; finally keeping the latest node and neighbor node information for the reliable transmit of the multicast video control information and recovering the loss locally if finding the control information loss. In the remaining part of the paper, the reliability recovery course and performance test are discussed. In the reliability recovery course, the related analyses are based on the probability theory of packet loss, in which node receiving probability and mechanism average receiving probability are definitely formulated. For the performance test, the related test course and test result analysis are given to show the reliable & scalable performance of the mechanism, in which the nodes simulation and loss simulation of the network are implemented by using the existed methods, which feasibility are be already validated. According to the theory analysis and experiment result, the hypercube-based reliable & scalable control shows efficient reliability and highly scalability, which performs more reliability with the increase of the nodes in the hypercube topology.