Improving data distribution in branching point based multicast protocols

Abstract

In several multicast routing protocols, the multicast distribution tree is identified by its branching points where multicast data is delivered from one branching point to another using native unicast. As a result, these protocols can be deployed incrementally and have low memory requirements compared to the traditional multicast routing protocols. The main drawback of these approaches is excessive lookup process in both multicast and unicast forwarding engines when handling data packets. To avoid this, we propose a minor modification in the parser code of routers and the IP header of the multicast data packets. Simulation results show that our techniques reduce the overall number of required lookups at least by 45.89% compared with non-optimized branching point based protocols like REUNITE [6] and HBH [7]. © Springer-Verlag Berlin Heidelberg 2004.