SUMMARYOwing to the wide deployment of broadband networks, information can now be rapidly disseminated all over the world. Only the content of static texts and pictures has not yet met users' demands. Along with the rapid advance of computer-related technologies, how to efficiently provide multimedia content for a large number of heterogeneous users on the Internet, especially videos, has become a noticeable issue. In this paper, we discuss the adaptation of multiple description coding (MDC) for the loopback buffering mechanism to address the issue of peer heterogeneity on content distribution network-peer to peer (CDN-P2P) video-on-demand systems. Three description selection policies, Ordered, Round-Robin, and OpenLoop-First (OLF), are studied. In addition, we propose the intra-description recovery and inter-description recovery to deal with failed peers. The simulation results show that the OLF with loopback-MDC can significantly reduce the amount of uploading bandwidth of a proxy on CDN-P2P video-on-demand systems, and the inter-description recovery, sewing, can further effectively recover missing blocks and restore the continuity of a breaking sharing loop among peers.

No abstract

SUMMARYIn a peer-to-peer (P2P) network, peers not only receive services from the network, but also contribute their own resources to the network. The abundant resources brought by P2P networks have stimulated the wide deployment of resource-consuming multimedia streaming services over a P2P architecture. Allowing for the heterogeneity of peers in their upload and download bandwidth, most researches adopt the Multiple Description Coding (MDC) technique to enable differentiated streaming services. However, due to the lack of a global view of description availability, they tend to cause the skewed distribution of descriptions and lead to the under-utilization of aggregated resources on P2P networks. In the paper, we propose a novel approach to balance the distribution of descriptions and available upload bandwidth in the P2P live streaming network. Based on the description distribution and available upload bandwidth, the proposed balancing scheme determines what descriptions a new peer should receive and their source peers such that better system scalability can be achieved. The simulation experiments indicate that our proposed balancing scheme can effectively reduce the server bandwidth consumption and rejection rate. Furthermore, abandoned peers can recover their lost descriptions mostly from the existing peers in the network instead of the server.

No abstract

In this paper, we present the performance result and analysis of parallel CBR video servers with conflict-free scheduling algorithms and server synchronization at time slot and time cycle levels. The experimental outcome shows that by relaxing serverlevel synchronization to a time cycle, the conflict-free slot scheduling produces higher stream throughput than the cycle scheduling.