[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [AVT] Three way and Conference calls
"Brian Rosen" <br at brianrosen.net> writes:
>RTP and RTPC are always used in a single end to end leg. The mixer has a
>point to point connection with each leg. The mixer can be in an endpoint or
>in a server. If a phone, for example, can mix a three way, it may simply
>have point to point RTP connections with the other two parties. If it
>can't, it may need an external mixer to do that. That mixer would have 3
>point to point RTP connections.
>
>
> B S
> / \ or / | \
> A C A B C
There's one other configuration for 3-way and conference calls:
B
/ \
A --- C
I.e. a no-central-mixer solution, or an "all to all media" topology as
defined in RFC 5117. Each endpoint would be responsible for doing it's own
mixing as needed, and the upstream and downstream bandwidth of A and C (for
the first diagram above) and A, B and C (for the second diagram) would be
roughly doubled compared to a mixer solution. A & C (or A, B and C) also
probably need to be capable enough to decode two incoming streams, though
they could implement a hard switch behavior (instead of mixing) and only
decode one stream at a time.
This configuration isn't often used in the unicast world, I believe; it
fits better in a multicast world where the upstream bandwidth of A & C (or
A, B & C) wouldn't have to increase. Also, as you add more participants,
the bandwidth and processing multiplication effect quickly will cause
problems.
>-----Original Message-----
>Plz provide info on Role of RTP/RTCP in three way and conference calls.
>During a conference call, RTP mixing has to be performed by the
>endpoints or servers. Plz clarify.
--
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)