Thesis (M.Sc.) - Cairo University - Faculty of Computers and Information - Department of Operations Research and Decision Support

It is well know that Peer- to- Peer (P2P) systems are advantageous in that the computational and bandwidth resources are distributed in a fashion that is more efficient as compared to systems that have central server and a multitude of clients. In the context of live streaming, there is a challenge in building systems that are both predictable and that realize minimum bandwidth load on the streaming server, i.e. maximum savings. The challenge is addressed by introducing a decentralized approach for peer placement and connectivity that are described as a probabilistic model. This approach can achieve bandwidth savings closely approaching what can be achieved in a centralized system with global knowledge. Peer placement is done by a decentralized overlay construction algorithm that manages to make use of an optimally generated centralized overlay construction algorithm

Issued also as CD