Overlay multicast solutions to media streaming

Media streaming is a core technology behind many important applications including digital libraries, distance learning, public information systems, movie on demand, and other corporate communications. This dissertation presents cost-effective and efficient solutions to implement this technology on the current Internet.; This implementation is not a trivial task. Indeed, media streams are usually voluminous in size and thus require huge network bandwidth when transmitted over communication networks. Furthermore, rendering a media stream at the receiver must follow stringent delay restrictions, making the delivery of this stream even more problematic. Although broadband networks can help speed up the transmission of media streams, designing a media streaming system that can scale well with the number of receivers remains challenging, mostly because of the limited capability of the media server. IP Multicast could be the best way to overcome this drawback since it was designed for group-oriented applications, but its deployment on the Internet has not been successful due to fundamental concerns.; Our objective is to seek scalable streaming solutions that employ IP Unicast only. Specifically, we investigate two overlay-multicast approaches to streaming media on the Internet. In the first approach, referred to as the overlay-router approach, a number of reliable servers are installed across the network to act as the software routers with multicast functionality. These application-level routers are interconnected according to a topology which forms an overlay for running the services. The content is transmitted from the source to a set of receivers on a multicast tree consisting of the overlay routers. A new receiver joins an existing media stream by connecting to an overlay router appearing on the delivery path to an existing receiver. This approach is scalable since the receivers can get the content not only from the source, but also from software routers, thus alleviating bandwidth demand at the source. The second overlay-multicast approach is the peer-to-peer approach. It assumes no extra resources such as the dedicated servers mentioned above. The multicast tree involves only the source and the receivers, thus avoiding the complexity and cost of deploying and maintaining extra servers.; This dissertation proposes two overlay-multicast schemes. The first scheme, called Range Multicast, is based on the overlay-router approach to solve the video-on-demand problem. The second scheme, called Zigzag, is based on the peer-to-peer approach to solve the live broadcast problem. We will show by in-depth performance studies that Range Multicast and Zigzag are truly promising solutions for large-scale video streaming applications on the Internet. The dissertation also describes a system prototype for Range Multicast and pointers to the future research.