Performance Comparison of LMS and PGM using Simulation

Most proposed reliable multicast schemes have difficulties satisfying all the requirements of emerging multicast applications. For example, SRM trades latency for reduced implosion, and lacks local recovery. Thus, recently, several proposals have been made for schemes that change the traditional IP Multicast model by requiring new services from routers. These router-assisted schemes promise great improvements in scalability and performance over traditional schemes. However, since most such schemes are still either paper designs or if implemented have not been evaluated, such promises have not yet been backed up. In this paper we take a step towards remedying that situation by using simulation to compare and evaluate two such schemes, namely Lightweight Multicast Services (LMS) and Pragmatic General Multicast (PGM), to determine how they perform against each other. Our results using the NS package show that despite its heavier network state, PGM is not as efficient as LMS due to repeated retransmissions and recovery from the sender. Our implementations of PGM and LMS will be publicly available to the NS community to be used along with the existing SRM implementation for comparing RM schemes.