Re: [AVT] RFC 3984 Packetization 0 question

[Randell Jesup <rjesup at wgate.com>]

Thomas Stockhammer <stockhammer at nomor.de> writes:
>FUs do have other use cases than the one you mention, for example if you
>transmit pre-encoded content and you have to adapt the packet size.
>In addition if you combine with Application-Layer FEC as for example the
>case in 3GPP MBMS, FUs adapted to radio packet payload sizes can improve
>performance significantly when compared to no adaptation or even when
>compared to the use of adapted NAL unit sizes (i.e. slices).

Right - and not just for pre-encoded content.

Think of an RTP gateway (translater, what have you) at the border of a
large-MTU network (ethernet, even jumbo-frame ethernet) and a small-MTU
network (think radio transmission such as 3G).  You can either adapt the
original encoder down to the small-MTU size (and hurt your compression
ratios, and increase number of packets on the large-MTU network a lot), or
you could have the gateway take large NALs and fragment them.  Effectively,
to the gateway, the incoming large NALs look like pre-encoded content.
This does require some fancy footwork and the gateway is really acting
(mostly) like a B2BUA for the RTP traffic.  It could do the same thing in
reverse (aggregate into STAPs) in the other direction.

Even in straight 1500-byte MTU networks, mode 1 should let large NALs get
better compression, though how much I don't know.

-- 
Randell Jesup, Worldgate (developers of the Ojo videophone), ex-Amiga OS team
rjesup at wgate.com
"The fetters imposed on liberty at home have ever been forged out of the weapons
provided for defence against real, pretended, or imaginary dangers from abroad."
		- James Madison, 4th US president (1751-1836)


_______________________________________________
Audio/Video Transport Working Group
avt at ietf.org
https://www1.ietf.org/mailman/listinfo/avt