[e2e] Protocols for tightly synchronised multicast?

Simon Spero ses at unc.edu
Tue Aug 14 18:36:48 PDT 2001


> Buffering is so cheap, why try to avoid it?  Just synchronize clocks at
all
> the receivers and transmit.

I think that's the right way of dealing with this; Buffering does add
latency, but I guess that's going to be swamped by  that coming from the
speed of sound;  no matter what, it's going to be too much for monitoring
sound -after that, the main constraint is keeping the latency low enough to
make it possible to monitor crowd reaction (if people start raising their
hands in the air 20 seconds after the breakdown ended, the bloke with the
Technics is going to get pretty confused pretty quickly.  I think the
buffering requirements for a small-diameter, mostly 802.11 based network
should be small enough to make this possible (though the downside of
wireless is that packet loss is often due to packets getting lost - that's
always so confusing).

I think I need to go find the back of an envelope :-)

Simon




More information about the end2end-interest mailing list