Peer-to-peer streaming based on network coding decreases packet jitter
Proceedings of the 2010 ACM workshop on Advanced video streaming techniques for peer-to-peer networks and social networking
| Hi-index | 0.00 |
Packet delay variation (or delay jitter) measurements are usedby applications to estimate the service quality received fromthe network, or by network operators to monitor network operationstates. Since a single jitter measurement takes twodelay samples to calculate, the time scale over which the twodelay samples are taken may affect the statistics of measuredjitter. The current common practice of calculating jitter statisticsis by treating all measurements as valid samples of thesame sampling space. In this paper, we perform scaling analyseson measured delay sequences to show that the properway of conducting jitter statistic analysis is by first groupingjitter samples into different clusters each containing samplesthat are taken over the same or similar time scales, andthen carrying out statistic analysis separately on these clusters.This special treatment is desired due to the existence ofstrong short-range dependency among packet delays, whichis introduced by queueing effect. The tool selected to performthe scaling analysis is called Deviation-Lag Function (DLF).We show that some congestion-related information of congestedend-to-end paths can be derived from their DLF plots.We also discuss the potential usage of DLF for bottleneckqueue detection.