Translating scalable video streams from wide-area to access networks

摘要

Transmitting video over user datagram protocol (UDP) has been considered advantageous because it allows for discarding of packets in favor of retransmissions, and sender-controlled timing. Using UDP has been criticized because it allows video streams to consume more than their fair share of bandwidth, which is typically associated with the back-off behavior of transmission control protocol (TCP). TCP-friendly algorithms are meant as a middle path. However, UDP delivery to end systems may still be prevented by firewalls or for other reasons, and TCP must be used. This in turn suffers from bandwidth fluctuations. Investigates an architecture that separates the transfer of a video stream over long distances. Considers a proxy server to translate the traffic and two straightforward approaches for the translation of a layered video stream transmission from the TCP-friendly transport protocol to TCP. Does not expect that one of these two approaches is by itself suited for the task, but investigating them will provide insights into their basic functions and help in discovering appropriate modifications. For the investigation, an experimental approach was used where network behavior and content are emulated.