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
One possibility is to reuse the TIAS definition from RFC 3890 section 6.2.2.
Note that this definition doesn't say what period it's calculated over, nor does it say how to deal with undershoots and overshoots. If the IETF can live without these parts of the definition, can the W3C do likewise?
Here is some proposed text based on RFC 3890 Section 6.2.2:
The Transport Independent Application Specific Maximum (TIAS) bandwidth .... is the maximum [amount of bandwidth needed when all declared media streams are used] ... without counting IP or other transport layers like TCP or UDP.... This can be less than the sum of all the individual media streams values... [since] not all streams have their maximum at the same point in time.
The discussion at TPAC and IETF invoked the question of what timeframe we should look for a bitrate to be satisfied over - one second was proposed, but seemed to be thought a bit arbitrary. (Obviously it's not smooth at timescales less than the video frame size or audio packet time.)
Left in the note about "TODO: Find or create a definition for bitrate (how much header overhead is included, for example). Should be aligned with RTCOutboundRTPStreamStats.targetBitrate in webrtc-stats.". Do we need another PR for targetBitrate?