|
|
|
Anirban Mahanti Department of Computer Science University of Saskatchewan Saskatoon, Saskatchewan |
|
In recent years, the Internet has experienced significant demand for multimedia applications that deliver stored media content on-demand to clients. Successful deployment of these applications has been limited, to some extent, by their high server and network bandwidth requirements, and by the lack of quality of service guarantees in the Internet.
Several scalable multicast protocols have been proposed that can stream popular content to a large numbers of clients with server and network bandwidth that grows sub-linearly with request rate. However, these protocols do not allow clients to recover from packet loss. This talk will outline the development of new periodic broadcast and bandwidth skimming protocols, namely Reliable Periodic Broadcast and Reliable Bandwidth Skimming, respectively that efficiently provide this capability. New scalability bounds will be discussed for both immediate service protocols as well as bounded delay protocols, assuming alternative packet loss recovery techniques. These bounds are used to guide the design and assess the performance of the new streaming protocols. The talk will outline how the new protocols can efficiently support streaming to heterogeneous clients, and conclude with directions for future work.