Research On Congestion Control Of Multi-source Application-layer Multicast

As an effective means of point-to-multipoint data transmission, multicast can effectively save network bandwidth and improve the efficiency of communication. However, IP multicast has not been extensively used by ISP, due to some difficult problems about congestion control and network deployment. Application layer multicast has solved the default that IP multicast had and provided a new solution for the group communicating application, which has gradually became the focus of network research. Its basic idea is to establish logical trees from sender to receiver among end-systems at application layer. It's to say that end-systems at application layer are in charge of multicast routing, and IP unicast transmission is still used in network layer. It not only avoids changing network center system and makes it easy to realize, but also can make full use of the storage capacity of end-systems at application layer, and solve problems such as reliable transmission with Qos machnism at IP layer or application layer. At present, most of the studies for application layer multicast congestion control only focus on one-source situation at home and abroad. So this paper designs a new kind of application layer multi-source multicast congestion control scheme based on the window, which is of significance for stableing performance of the whole network and improving the throughput of multicast.This paper designed an application layer multi-source multicast congestion control scheme based on the window; the plan includes three models: the window back pressure mechanism, transfer in nodes and congestion information feedback. Also realized OMCC- the prototype system of above solution at PlanetLab- a simulation platform; took experiments and analyzed the performance about efficiency, fairness and scalability; the experimental results show that this scheme is efficient at congestion control, realized the reliability of data transmission and achieved fairness of multi-source. Finally summarized the work and shortcomings in the paper, and point out the direction of further research.