Modeling Up the Network Synchronization Problem - 10/28/2011

Jiaxi Jin

Title: Modeling Up the Network Synchronization Problem

Abstract: Assume that there are a number of distributed and synchronized databases exist in a connected network, at a moment new update is made on some of the databases, the problem of making the databases in the whole network become synchronized (again) is so called network synchronization problem. The network with synchronization problem can be mapped to a graph, named network graph. In a network graph, each vertex represents one database in the network, and has three possible colors representing its state: i) white (unsynchronized), whose database doesn’t contain any new updates; ii) gray (partially synchronized), whose database contains some of the new update, but not complete; or iii) black (fully synchronized), whose contains all the new update completely. An edge exists between two vertices iff a communication channel exists between the two represented devices.

The actual proceeding of network synchronization varies from network topologies, protocols deployed, initial conditions, etc. In order to carry on performance analysis on network synchronization problem, an appropriate model with clear mathematical description on the synchronization process is needed. In this talk, the speaker will introduce and evaluate different ways of establishing such a model which might be used in network synchronization analysis.

-- JiaxiJin - 09 Dec 2011