08-16-2017, 10:37 PM
Broadcasting and gossiping are fundamental tasks in network communication. As communication networks grow in size, they become increasingly vulnerable to component failures. Some links and/or nodes of the network may fail. It becomes important to design communication algorithms in such a way that the desired communication task be accomplished efficiently in spite of these faults, usually without knowing their location ahead of time.
In this seminars, we can review the history of Fault-Tolerant broadcasting and gossiping research, and present you several existing Fault-Tolerant algorithms. Here, we consider two alternative assumptions concerning fault distribution. The bounded fault model assumes an upper bound on the number of faults and their worst-case location, while in the probabilistic model faults are supposed random and independent. Faults are assumed either of crash type (a faulty page link or node does not transmit) or of Byzantine type (a faulty page link or node may corrupt transmitted messages).