Skip to Main Content
Peer-to-peer (P2P) networks represent a valuable architecture for streaming video over the Internet. In these systems, users contribute their resources to relay the media to others and no dedicated infrastructure is required. In order to ensure a low end-to-end delay, P2P overlay networks are often organized as a set of complementary multicast trees. The source of the stream multiplexes the data on top of these trees and the routing of packets is statically defined. In this scenario, the reliability of the overlay links is critical for the performance of the system since temporary link failure or network congestion can cause a significant disruption of the end-user quality. The novel Scalable Video Coding (SVC) standard enables efficient usage of the network capacity by allowing intermediate high capacity nodes in the overlay network to dynamically extract layers from the scalable bit stream to serve less capable peers. On the other hand, SVC incurs a certain loss in terms of coding efficiency with respect to H.264/AVC single-layer coding. We propose a simple model that allows to evaluate the trade-off of using a scalable codec with respect to single-layer coding, given the distribution of the receivers’ capacities in an error-free network. We also report experimental results obtained by using SVC on top of a real-time implementation of the Stanford Peer-to-Peer Multicast (SPPM) protocol that clearly show the benefits of a prioritization mechanism to react to network congestion.