Research On Application-Level Multicast Models For Large Scale Live Media Streaming Service

Nowadays, multmedia streaming over the Internet is booming, However, its QoS is generally unsatisfactory. Traditional Client/Server model isn't able to be adopted to large-scale distubution of media streaming because of Servers' performance bottleneck. And IP multicast has not been widely deployed because of poor scalability, unmanageability, complexity of group address management and dependency to specific multicast infrastructure. At present, Application-Level Multicast(ALM) becomes a hot research in Media Streaming Distribution. However, the existing ALM is difficult to meet the need of large scale live media streaming service in scalable,reliable and real time transmission.A new ALM model HCALM is proposed which combines ALM based on Gossip with Client/Server model, for efficiently supports the distribution of large scale live media streaming.1,A Hierachical-Cluster P2P Overlay Network HCALM is proposed as the infrastructure of application of relevant key technology, which can handle the heterogeneity of node capacity.2,A self-organized, self-adapted mechanism for node organization and management is designed, which constructs the overlay network in a distributed,efficient scalable manner.3,A hybrid optimization topology strategy HOA is proposed according to HCALM, which can optimize overlay structure from different levels in different ways.4,A Subscription-based Multi-Sender data transmission strategy is proposed which greatly decreases the latency.