Evaluating Fault-Tolerance Algorithms for Multicast Services

In many distributed systems, reliable point-to-point communication, such as D2D (device-to-device) communication, and MANET (Mobile Ad-Hoc Network), is established by making use of a reliable transport protocol. Considering how important point-to-point communication is, it is not surprising that reliable multicast services are important as well. Although most transport layers offer reliable point-to-point communication, they rarely offer reliable channels to a collection of processes. In most cases, achieving reliable communication comes at a cost of overhead. Overhead messages in MANET grow proportionally with the frequency of changes in topology. Networks are becoming more dynamic, and the overhead incurred is limiting the possible MCR (Message Completion Ratio) and performance of current networks. In this work, we investigate different approaches for fault-tolerance. Methods range from standard fault-tolerance algorithms, such as TCP, where the focus is on unicast services, as opposed to SRM (Scalable Reliable Multicast), a non-standard fault-tolerance algorithm for multicast services. The trade-off between MCR and overhead is shown through comparative simulations.