Data distribution over an overlay network

The Client-Server model based data distribution is inefficient for sessions with a large number of participants interested in receiving the same content at the same instant. Examples of such applications are live audio/video streaming, weather updates, stock tickers etc. The lack of global multicast infrastructure has made the research community to consider 'Overlay networks' as alternatives. Overlay networks require effective mechanisms for bootstrapping, constructing, maintaining and repairing the overlay. The effectiveness of these mechanisms influences the quality of the service experienced using the overlay network. In this thesis, we propose solutions that can be used by the overlay network to construct, maintain and repair the overlay. More precisely, the solutions that we propose can construct a minimum spanning tree for data distribution and identify capable (nodes with extra outbound degree) nodes using a decentralised design. Overlay networks can be classified into different types depending on the nature of the participants and the type of data distribution mechanism (tree, mesh). In the thesis, our focus is only on the overlay networks that uses tree based data distribution mechanism