A reliable multicast protocol based on multiple multicast groups and error-recovery channel

We examine an approach for providing scalable reliable multicast service using multiple multicast groups and error-recovery channel. In this approach, receivers are first divided into multiple multicast groups. In each of these groups, the processing of acknowledgement and retransmission is assigned to its group manager. So, the processing workload is well distributed among participants in a balanced way, and the packet traffic flowing into backbone network is efficiently decreased. At the same time, rather than having all receivers receive all retransmitted packets, error-recovery channel, a special multicast channel for retransmission, is introduced to allow only those receivers that actually want a particular packet to actually receive that packet. This effectively reduces receivers' processing cost and bandwidth consumption because receivers do not receive the retransmission of packets they have received correctly. Finally, the performance analysis of throughput and packet traffic on backbone network is conducted to evaluate the protocol. Numerical results show that the proposed protocol makes efficient use of both network and end-host resources, and scales to large-scale one-to-many and many-to-many collaborative applications that can potentially have several thousand users.