| Home
Papers
Download
|
Probabilistic Congestion Control
Probabilistic Congestion Control (PCC) is a TCP-friendly congestion
control scheme for non-adaptable flows. The main characteristic of
these flows is that their data rate is determined by an application
and cannot be adapted to the current congestion situation of the
network. Typical examples of non-adaptable flows are those produced
by networked computer games or live audio and video transmissions
where adaption of the quality is not possible (e.g., since it is
already at the lowest possible quality). PCC performs
congestion control for non-adaptable flows by suspending them at
appropriate times so that the aggregation of multiple non-adaptable
flows behaves in a TCP-friendly manner. The decision whether a flow
is to be suspended is based on random experiments. In order to
allocate probabilities for these experiments, the data rate of the
non-adaptable flow is compared to the rate that a TCP flow would
achieve under the same conditions.
For questions regarding PCC contact Martin Mauve and Jörg Widmer.
|