A simulation framework for the analysis of multicast tree algorithms

Group communications in high speed networks require the development of efficient multicast algorithms. Finding the optimal multicast routing tree for a subset of a network nodes is an NP-complete problem known as the Steiner Minimal Tree (SMT). Several heuristics were developed to provide an approximate solutions for this problem. The analysis of these heuristics, however, have been limited to specific network topologies. This paper discusses a flexible simulation framework to study the performance of multicasting algorithms in different topologies, including dense and sparse networks. The framework is then used to provide a detailed analysis of the performance of a selected set of path distance heuristics frequently discussed in the literature. The performance of these heuristics as then compared to two new heuristics, namely Normalized Average Distance Heuristic (NADH) and Shared Average Distance Heuristic (SADH). The results show that, on average, NADH outperforms all other heuristics in dense network topologies. The results also show that SADH out-performs the selected set of path distance heuristics for most network topologies.