Logo image
The Design and Implementation of Peer Structure for a P2P Video Streaming Delivering System with Guaranteed Latency
Thesis

The Design and Implementation of Peer Structure for a P2P Video Streaming Delivering System with Guaranteed Latency

Li, Zong-Yo
Masters, 國立清華大學, 通訊工程研究所
2012

Abstract

點對點 串流 網狀拓樸 樹狀拓樸 低延遲 現場轉播 p2p streaming mesh topology tree topology low latency live broadcast
Abstract Peer-to-Peer (P2P) network applications have become increasingly widespread in today’s Internet. The primary reason for the emergence of P2P network has been the ability to reduce server burdens and bandwidth costs. However, these benefits do not automatically guarantee a reduction in, and/or control of, transmission delays. When streaming live broadcasts, transmission delays are an important factor directly affecting the quality of user experience. Generally speaking, in a large-scale live broadcast using P2P technology, minimizing transmission delays is not crucial as long as the latency is consistent and stable; however, for an online video conferencing system, transmission delays must be strictly controlled so as to be no more than a few seconds, in order to ensure the interactivity among the conference participants. This paper proposes a P2P system using a hybrid mesh topology. We call this system “Low-Latency Peer-to-Peer Network” (LLP2P). One of the key features of this system is the ability to minimize server output bandwidth at any specified length of latency. In addition, this system also provides a series of detection and rescue algorithms as mechanisms that allow users to quickly recover from poor streaming performance caused by unstable network quality for their parent nodes. With these mechanisms, we can ensure that the streaming latency for every user is below our specified value, so as to provide the same quality service, in a sustainable manner, as would be offered by the client-server architecture. We have also designed various experiments to test our system, including joining and leaving nodes, detection algorithm, and rescue mechanism. Data from these experiments have been collected and statistically analyzed, in order to verify the feasibility of our system accordingly.

Metrics

1 Record Views

Details

Logo image