H264 Packing Mode for FUA

Question

We have a couple of interaction questions, where: The video mode that is required for a pair of endpoints on the market is slightly different and only understands the H.264 packetization modes (FUA type) (i.e. FU -A NAL type). (while others do not play the video when receiving a payload like fu-nal)

Does anyone know what kind of FUA packaging is this? How is it different from the 0.1,2 batch modes as defined in RFC3984? Does it support a video encoder / decoder, how can it be appropriately signaled in a SIP SDP session in which attributes do not change even when passing through SIP B2BUAs?

Answer 1

FUA is used in packetization modes 1 and 2. By default, packetization mode is 0 (single NAL mode); if both sides agree on modes 1 or 2, you will usually see FU-A when the NAL exceeds the UDP MTU or the configured maximum packet size.

The packetization / depacketization level should receive NAL and, if necessary, generate FU-A, and upon reception, accept the FU-A series and collect the NAL for delivery to the decoder.

See RFC 3984 and RFC 3984bis (of which I am the author).

A B2BUA should not accept or offer a packetization mode that it is not ready for processing, although it may simply go through a proposal / response from another UA.

Also note that (if it is compatible with RFC 3984), the UA must support packetization mode 0, although it should not offer it on INVITE.

Answer 2

I'm not sure if I understood your question correctly, but FU-A is not a packetization mode, it just signals that the NAL block is a Nal fragmentation block, that is, the NAL block is fragmented by several RTP packets. Table 3 of RFC3984 shows that NAL FU-A can only be used in alternating and non-moving packetization modes (modes 1 and 2), that is, in more than one Nal Unit mode (mode 0).

As for the support of the encoder / decoder: if the SDP signals packetization mode 1 or 2, this means that there may be FU-As in the RTP stream. This will not affect the decoder, although RFC3894 section 7.1: "If the decapsulated packet is FU-A, all fragments of the fragmented NAL unit are combined and transmitted to the decoder."

I don’t understand the last part of your question regarding SIP B2BUAs, SDP signals packetization modes and on the basis that the RTP receiver should be able to handle the various types of NAL elements specified in RFC3984.