You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
"A desire has been stated for integrating congestion control across SCTP and media, using a realtime congestion control protocol like Google Transport-wide-CC, SCREAM or NADA. There's some wishful thinking in -rtcweb-transport about how such a combo should work, but I think there are probably dragons once we try implementing it."
The text was updated successfully, but these errors were encountered:
Some recent discussion of the interactions between an application looking to achieve low-latency in media transport and an underlying transport implementing NewReno is here.
While certain combinations of app-layer congestion control and transport-layer congestion control may appear to be compatible in simulations, this is likely to be fragile in practice. For example, a media application may not be able to instantaneously adjust its sending rate (e.g. by dropping or adding layers), but may need to wait to adjust the average bitrate target at a keyframe. Also, low-latency CC algorithms may need one-way delay info that may require extensions to the transport as well as statistics.
Discussed at TPAC 2022. Consensus seems that this is 1) hard to do and 2) not something that needs to show up at the Web API anyway (quality of implementation and/or IETF issue).
From Harald (Issue 71):
"A desire has been stated for integrating congestion control across SCTP and media, using a realtime congestion control protocol like Google Transport-wide-CC, SCREAM or NADA. There's some wishful thinking in -rtcweb-transport about how such a combo should work, but I think there are probably dragons once we try implementing it."
The text was updated successfully, but these errors were encountered: