[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

RE: [AVT] RTP streaming and adaptation to AVC of an SVC temporalscalable bitstream - Packet loss

Hi Daniele,
 
In your example the SVCtoAVC adapter is an RTP mixer, which terminates the RTP session between the sender and itself and restarts another RTP session between itself and the receiver. Therefore the RTP sequence number needs to be updated for the base layer packets anyway.
 
BR, YK

From: ext daniele renzi (bsoft) [mailto:daniele at bsoft.info]
Sent: Friday, August 17, 2007 5:41 PM
To: avt at ietf.org
Subject: [AVT] RTP streaming and adaptation to AVC of an SVC temporalscalable bitstream - Packet loss

Dear all,
 
I'd like to clarify the following scenario (the reference documents are draft-ietf-avt-rtp-svc-02.txt and RFC-3984).
 
We're implementing a system including:
a) an SVC encoder + RTP streamer module (non-interleaved mode),
b) an SVCtoAVC adapter, taking the RTP stream from the network, adapting its SVC payload to AVC, streaming (RTP) the resulting AVC stream,
c) an RTP receiver + AVC decoder.
 
The SVC encoder implements just temporal scalability: it produces/sends AVC-compliant NALUs and before any of them a prefix NALU to carry scalability information.
The different (temporal) layers are transported in a single RTP session.
 
Let's suppose that the SVCtoAVC adapter has to forward just the base layer by dropping the enhancement layers and the prefix NALUs and making just some minor adaptation in the Sequence Parameter Set.
 
Now the question: how can we handle a packet loss in the path between the SVC encoder and the SVCtoAVC adapter, considering the single RTP session?
The problem is that the SVCtoAVC adapter is not able to know which layer the lost packet belongs to; it can just detect a generic packet loss, which could affect either the base layer or an enhancement layer, then the SVCtoAVC adapter doesn't know whether this loss has to be signaled to the receiver, i.e. whether it must insert a sequence number gap in the outcoming RTP stream...
 
If we used different RTP sessions the hardle would be cleared, as the sequence number would be increased separately for any layer. Anyway, we'd need to keep the single RTP session mode (if allowed, of course).
 
Any suggestion would be very appreciated.
 
Thanks in advance.
 
Best regards,
 
 
Daniele Renzi
bSoft -- www.bsoft.info
+39-0733-57707  (tel/fax)
_______________________________________________
Audio/Video Transport Working Group
avt at ietf.org
https://www1.ietf.org/mailman/listinfo/avt