首页> 外文会议>2015 2nd World Symposium on Web Applications and Networking >Efficient multi-source RTP stream relaying in overlay networks
【24h】

Efficient multi-source RTP stream relaying in overlay networks

机译:覆盖网络中的高效多源RTP流中继

获取原文
获取原文并翻译 | 示例

摘要

P2P streaming systems exploit the high scalability of P2P networks to propagate multimedia streams to wide areas in a distributed fashion. A P2P network is made up of several logic connections among peer nodes, which can form a mesh or a tree overlay. Interactive real-time streaming imposes very strict delay constraints: in e-learning environments, for instance, only a high level of synchrony with the live event may allow users to ask questions in real-time to the lecture speaker. Despite their resilience to peer churning, mesh overlays are affected by a high and variable delay, which hampers their adoption in interactive applications. Playback discontinuity, caused by missing chunks, may be a serious problem too, since it may compromise the speech understanding. On the contrary, multi-tree MDC-based overlays can guarantee a smooth playback continuity, since a stream can be fluently reproduced also when some descriptions are lost. The queuing and the elaboration delay of the nodes should be kept under control to reduce the overall end-to-end delay on each tree branch. In this work we describe a framework for the construction of reliable tree overlays: we designed a smooth switch-to-fallback mechanism to quickly replace nodes leaving the overlay, a channel for packet recovery and a very efficient engine that collects MDC descriptions coming from parent nodes. The experimental data we reported in the last section reveal a low CPU usage and a very low elaboration delay in collecting MDC descriptions.
机译:P2P流传输系统利用P2P网络的高可伸缩性,以分布式方式将多媒体流传播到广泛的区域。 P2P网络由对等节点之间的几个逻辑连接组成,这些逻辑连接可以形成网格或树状覆盖图。交互式实时流强加了非常严格的延迟约束:例如,在电子学习环境中,只有与实时事件的高度同步才能允许用户向演讲者实时提问。尽管网格覆盖可以抵抗同伴搅动,但网格覆盖仍会受到高延迟和可变延迟的影响,这会阻碍网格覆盖在交互式应用程序中的采用。由于缺少块而导致的播放不连续性,也可能是一个严重的问题,因为它可能会损害语音理解能力。相反,基于多树基于MDC的覆盖可以保证平滑的播放连续性,因为在丢失某些描述时也可以流畅地再现流。节点的排队和精心设计的延迟应保持在控制之下,以减少每个树枝上总的端到端延迟。在这项工作中,我们描述了一个用于构建可靠树覆盖的框架:我们设计了一种平滑的切换到后备机制,以快速替换离开覆盖的节点,一个用于数据包恢复的通道以及一个非常有效的引擎,该引擎可以收集来自父节点的MDC描述节点。我们在上一节中报告的实验数据显示,在收集MDC描述时,CPU使用率较低,并且设计延迟很低。

著录项

相似文献

  • 外文文献
  • 中文文献
  • 专利
获取原文

客服邮箱:kefu@zhangqiaokeyan.com

京公网安备:11010802029741号 ICP备案号:京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有
  • 客服微信

  • 服务号