Opportunistic native multicast under churn

Abstract

In this paper we investigate the behaviour of Opportunistic Native Multicast under node churn. Previously, we have introduced an Opportunistic Native Multicast approach that facilitates native multicast where possible and which reverts to using overlay multicast when needed. Here, we are evaluating Overlay Native Multicast in a more realistic environment considering the effect of network churn and propose election algorithms to offer resilience and adaptivity required in the real world. We have developed a mechanism where the network can autonomously and deterministically elect a primary node for each multicast island. We have further investigated ways to improve the efficiency and responsiveness of the protocol by introducing a secondary node selection process. We have tested our proposed improvements using Oversim simulation. We show that Opportunistic Native Multicast can achieve success rates well above 90 percent despite realistic levels of node churn.