Download presentation
Presentation is loading. Please wait.
Published byJohn Neal Modified over 9 years ago
1
Slide title minimum 48 pt Slide subtitle minimum 30 pt RTP Multiple Stream Sessions and Simulcast draft-westerlund-avtcore-multistream-and-simulcast-00 Magnus Westerlund and Bo Burman Ericsson
2
draft-westerlund-avtcore-multistream-and-simulcast-00 | IETF 81 | 2011-07-14 | Page 2 IPR Disclosure ›Telefonaktiebolaget LM Ericsson (publ)'s has made a Statement about IPR related to this draft in https://datatracker.ietf.org/ipr/1592/ https://datatracker.ietf.org/ipr/1592/
3
draft-westerlund-avtcore-multistream-and-simulcast-00 | IETF 81 | 2011-07-14 | Page 3 Outline ›Problem Descriptions –A Target Scenario –Simulcast –Multiple Streams in Advanced RTP usages –Bandwidth Signaling –Codec Control / Optimization ›Problem Summary ›Extensions –Multiple Streams signaling –Bandwidth SDP attribute –Simulcast Grouping –SRCNAME SDES item –Codec Control ›Way Forward –Architecture document –Extensions
4
draft-westerlund-avtcore-multistream-and-simulcast-00 | IETF 81 | 2011-07-14 | Page 4 A Target Scenario Active speaker RTP Mixer Listener …
5
draft-westerlund-avtcore-multistream-and-simulcast-00 | IETF 81 | 2011-07-14 | Page 5 Simulcast RTP Mixer Active speaker Listener When using scalable coding with inter-layer prediction, the broad arrows will be slightly smaller, but all red arrows will also have to be sent Simulcast: sending multiple representations of the same source
6
draft-westerlund-avtcore-multistream-and-simulcast-00 | IETF 81 | 2011-07-14 | Page 6 Simulcast and Scalable Encoding ›Simulcast is both an alternative and complementary to Scalable Encoding ›The trade-offs when it comes to efficiency are different –SVC encoding is more efficient in sender to mixer path –Simulcast is more efficient in mixer to receiver path –Combining scalable encoding with simulcast for best of both worlds ›Simulcast is codec agnostic ›Simulcast can be done for other purposes –Provide two different encodings for interoperability –Provide redundancy for robustness
7
draft-westerlund-avtcore-multistream-and-simulcast-00 | IETF 81 | 2011-07-14 | Page 7 Multiple Streams RTP Endpoint A sample client, both sending and receiving multiple video streams
8
draft-westerlund-avtcore-multistream-and-simulcast-00 | IETF 81 | 2011-07-14 | Page 8 Multi-Stream Layering ›An RTP Session can contain 1..N SSRCs ›An RTP Session is identified by a lower layer identifier, such as a UDP port or five tuples ›A multimedia session contains one or more RTP sessions UDP RTP Session SSRC 1 … SSRC 2 SSRC N RTP Session SSRC 1 … SSRC 2 SSRC N ENCDEC ENCDECENC
9
draft-westerlund-avtcore-multistream-and-simulcast-00 | IETF 81 | 2011-07-14 | Page 9 Multi-Stream Simulcast Layering ›A single source can be simulcasted as different versions –Same actual source in several sessions; don’t want to force new semantics into SSRC value ›Several sources can be rendered at the same device UDP RTP Session (Content: Main) SSRC 1 … SSRC 2 SSRC N RTP Session (Content: Alternative) SSRC 1 … SSRC 2 SSRC N ENCDEC ENCDEC Scene Generator
10
draft-westerlund-avtcore-multistream-and-simulcast-00 | IETF 81 | 2011-07-14 | Page 10 Multi-Stream Issues ›More advanced use cases than point to point VoIP: –Video conferencing –Telepresence –IPTV –Etc. ›This can result in multiple media streams –Is the end-point capable of handling multiple simultaneous media streams of the same media type? ›Legacy capabilities is likely one SSRC per direction –When should additional media streams be in the same RTP session, when in a new session? –When streams have relations, how to express that for: ›Retransmission ›Redundancy ›Simulcast
11
draft-westerlund-avtcore-multistream-and-simulcast-00 | IETF 81 | 2011-07-14 | Page 11 Bandwidth Signaling ›Current SDP Bandwidth signaling insufficient in handling: –Asymmetric bandwidth capabilities in the path –Asymmetric bandwidth usage inherent from application –When different Payload Types have different bandwidth ranges –When multi-stream applications use multiple streams in each direction –The allowed burstiness of media sources is not explicit pt=96 End- point Alternate codecs pt=8 End- point VGA HD
12
draft-westerlund-avtcore-multistream-and-simulcast-00 | IETF 81 | 2011-07-14 | Page 12 Codec Control / Optimization RTP Mixer Active speaker Listener Simulcast NOTE! Just as valid for scalable coding! Red, sent streams are not consumed by anyone (e.g. based on speaker activity) and should be paused and resumed on request
13
draft-westerlund-avtcore-multistream-and-simulcast-00 | IETF 81 | 2011-07-14 | Page 13 Problem Summary ›We have a general RTP architecture clarity issue –We need to clarify multiple SSRCs in one RTP session –We need to discuss when appropriate to use multiple RTP sessions –We need to create common principles for streams that aren’t independent, but have common source. ›We need to do this for all reasonable topologies ›Multiple SSRCs in an RTP session appear to need signaling support to avoid legacy issues ›Simulcast is good tool, we need signaling and RTP association mechanisms to make it work ›Bandwidth configuration and capability declaration in asymmetric usages and encodings needs to be improved ›Scalable Codecs and Simulcast needs additional Codec Control tools to optimize sessions
14
draft-westerlund-avtcore-multistream-and-simulcast-00 | IETF 81 | 2011-07-14 | Page 14 Proposed Extensions (1/4) ›Multiple Streams Signaling –Separated directions ›a=max-send-ssrc:96 2 ›a=max-recv-ssrc:96 5 –Both payload specific and payload agnostic ›a=max-recv-ssrc:98 6 ›a=max-recv-ssrc:99 4 ›a=max-recv-ssrc:* 8
15
draft-westerlund-avtcore-multistream-and-simulcast-00 | IETF 81 | 2011-07-14 | Page 15 Proposed Extensions (2/4) ›Bandwidth Signaling –b= line not possible to extend with sufficient new semantics –Per direction and payload type (also payload agnostic) –Per source ›a=bw:recv pt=96 SMT:tb=64000:320 ›a=bw:recv pt=97 SMT:tb=12200:128 –Entire media level aggregate ›a=bw:send pt=* AMT:tb=384000:512 –Allow for future needed semantics to be defined
16
draft-westerlund-avtcore-multistream-and-simulcast-00 | IETF 81 | 2011-07-14 | Page 16 Proposed Extensions (3/4) ›Simulcast Grouping in SDP –Different semantics between directions –a=group:SCS 1 2 3 … (SimulCast Send intention) –a=group:SCR 4 5 … (SimulCast Receive capability / acknowledge) ›Simulcast Source Identification in RTP –SDES CNAME is defined as unique per endpoint, not per source –New SDES SRCNAME unique per actual media source ›Indicate which streams are alternative encodings to each other
17
draft-westerlund-avtcore-multistream-and-simulcast-00 | IETF 81 | 2011-07-14 | Page 17 Proposed Extensions (4/4) ›Codec Control extensions is forthcomming
18
draft-westerlund-avtcore-multistream-and-simulcast-00 | IETF 81 | 2011-07-14 | Page 18 Anticipated Document Split Current Multi- stream Signalling Simulcast Grouping Bandwidth Codec Control RTP Session and Multi- stream Architecture
19
draft-westerlund-avtcore-multistream-and-simulcast-00 | IETF 81 | 2011-07-14 | Page 19 Proposal for Going Forward ›That AVTCORE takes on the general Architecture questions: –Make it clear when appropriate to use multiple streams within an RTP session –How should one use RTP sessions and SSRCs when having alternative, complementary or redundant streams ›That the various extensions are submitted to the appropriate WG as individual pieces for progressing: –AVTCORE: ›Architecture ›Multi-stream Signaling –AVTEXT: ›Simulcast Group Signaling ›Codec Control –MMUSIC: ›Bandwidth Signaling
Similar presentations
© 2025 SlidePlayer.com. Inc.
All rights reserved.