Performance Model of Reliable Hybrid Multicast Protocol (RHMP)

Abstract

The Pragmatic General multicast (PGM) and Elastic Reliable Multicast (ERM) are reliable multicast protocols. The difference between reliable and unreliable multicast protocol is that they make sure that the multicast data packets gets to its destinations. Both the PGM and ERM sends flood messages to the Rendezvous Point source (RPS) from the source node towards the stub nodes which then forward it to leaf nodes, leaf nodes that are not interested sends a prune message while any leaf node that misses a packet sends a message to the RPS through the stub node requesting for the multicast packet. A repair multicast packet is then forwarded to all leaf nodes that requested for it. In the reliable hybrid  multicast protocol (RHMP) being proposed the stub nodes originates the flood message to the leaf and uninterested leaf sends prune message, any stub that has one or more interested leaf sends a join message to the RPS. If a leaf node in the multicast distribution misses a multicast packet it requests a repair packet from its stub node and the stub node sends the repair data. A simulation model was developed to mimic the behaviour of PGM, ERM and RHMP in different network size using hierarchical network and the control bandwidth overhead (CBO) for each of the multicast protocols was calculated, the CBO was use as the cost metric. The result shows that the RHMP uses less CBO than PGM and ERM in a sparsely and densely populated network. For state storage it was discovered that the RHMP uses more resources at the stub nodes than at the source / RPS or leaf node when compared with PGM and ERM, but since the stub nodes are present in a distributed way it does not necessarily affect the multicast process.