| |
Based
on TCP/IP multicast theory, using a common class D multicast
group address, an initiating multicast system (Chairman)
is able to call multiple endpoints and broadcast audio
and video data to them simultaneously. Unlike unicast
transmission, which would require central MCU hardware
or software, multicast operates with single streams
of video and audio, reducing the overall bandwidth required
for the conference.
In a unicast call, between two endpoints
on the network, the amount of bandwidth utilised must
be doubled as there is video and audio data being transmitted
by both endpoints simultaneously. If this conference
is expanded (through centralised MCU hardware or software),
the bandwidth used on the network increases by a factor
of 2 for each endpoint added to the conference.
With multicast, the video and audio
data is a one-way transmission, and therefore, the overall
bandwidth of the conference never changes, regardless
of the number of endpoints that join.

Interactive Multicast takes this
theory one step further - although multicast is very
useful for one-way communication, by adding small, direct
unicast channels between the receiving stations and
the chairman, it is possible to send requests and messages
between participants. This communication 'back channel'
thus provides us with a control method for switching
off the multicast at one station and switching it on
at another, effectively switching the source of transmission
amongst participants as required. |
|