Design And Research Of StreamCast-A Streaming Media Distribution System Based On Peer-to-Peer Network

Due to the sparese deployment of IP multicast in the Internet today, someresearchers have proposed application layer multicast as a new approach to implementwidearea multicast services. Application layer multicast protocols organize a set ofhosts into an overlay tree for date delivery. Each end host on the overlay peers with asubset of other hosts. Unlike network-layer multicast, application layer multicastrequires no infrastructure support and can be easily deployed in the Internet.This dissertation first discusses the evolution trends and research status of P2Pnetwork and media streaming. Then we describe a set of application layer multicastprotocols that have been proposed in recent literature, classify them based on someproperties and present a comparision of performance and applicability of theseschemes.In this dissertation, we describe a brand new scalable application-layer multicastprotocol, StreamCast, specifically designed for media streaming applications withlarge receiver sets. Our scheme is based upon a hierachical clustering of theapplication-layer multicast peers and can support a number of different data deliverytrees with desirable properties.Since application layer multicast is implement at the end hosts instead ofnetwork routers and relies only on an underlying unicast architecture, multiple copiesof the same packet can be carried by a single physical link or node on the overlay,such system is undoubtedly less efficient than IP multicast. This dissertation study theStreamCast application layer multicast protocol presented before to quantify andstudy the tradeoff between two important metrics – stress and stretch in scalablybuilding application layer multicast paths.In the fifth part of this dissertation, we present extensive simulations of both ourprotocol and the Narada application layer multicast protocol over the Internet-liketopologies. Our results show that for groups of size 32 or more, our protocol has lowerlink stress (by about 25%), improved or similar end-to-end latencies and similarfailure recovery properties. More importantly, it is able to achieve these results byusing orders of magnitude lower control traffic.Finally we expect the study of StreamCast. As a future work, we will improvethe StreamCast protocol as the enhancement scheme mentioned in 3.5. Another workdirection is to further refurbish the analysis model and simulation tools. Furthermore,we can partially implement StreamCast protocol and experiment with it in the realnetwork environment.