The Problem Of Multi-rate Multicast Rate Allocation Based On Overlay Network

Multicast is an important research topic for the Internet. Due to technical and market problems of IP multicast brought about, IP multicast has not been deployed large-scale. Application layer multicast (ALM) was proposed, which was based on IP multicast model, Overlay network and Peer-to-Peer technologies. ALM become a new strategy that is the deployable alternative to IP multicast for the Internet.In ALM , ALM protocols organize a set of end hosts into an overlay tree for data delivery without supporting of network layer. Althrough it doesn't more efficient than IP multicast, it has significant practical sense because it doesn't need change exist network, and routing path can dynamicly adjust with the change of network. Appliance program can also take part in establishment of routing strategy, and ALM can achieve strong flexible that IP multicast couldn't achieve. Now there are many stream appliances based on overlay multicast in Internet.Multirate multicast can accommodate heterogeneity of Internet, and efficiently utilize the network resource. It is a good scheme for controlling network congestion. We introduced multirate multicast into overlay network. To the question of data constraint in overlay multicast, this paper focused on streaming of overlay multicast, and probed into rate allocation of streaming multicast based on analyzed overlay multicast.The chief of this paper contains the following four issues:1. Aimed at any overlay multicast tree, we proposed a rate allocation algorithm based on receivers under max-min fairness. With this algorithm, the network resource utility is optimal, ie.utility of network is maximal while achieving maximal minimum rate under max-min fairness.2. In order to solve the NP hard problem of constructing overlay multicast trees, a multicast tree heuristic algorithm, based on the degree constraint of nodes, is proposed. In fact, the multicast tree that constructed by the new algorithm is a hypo-optimal tree.3. We have run the simulation in OPNET modeler, and put max-min rate allocation in practice on the tree that constructed by the heuristic. We compaired the heuristic with max-min tree and max-throughput tree. It was proved that the performances of the heuristic was between the other two optimal trees above, and had higher link utilization and throughput.The contribution of this paper is that applying the heuristic to construct overlay multicast tree with optimal in rate allocation, and supporting fair multirate overlay multicast in thin granularity.