Erlang?

[Bob Bruhns]

The Erlang idea is just an application of statistics.  If you can
assume a typical probability distribution and account for typical
variations over time and circumstance, then you can reasonably
quantify the likelihood of collision or clear sailing in a shared
system at any given moment for any given connection path.  If other
things happen, such as a flood of retries after a normal timeout has
elapsed, that just introduces a nonlinearity with a time-delay.
Some reasonable assumptions can be made, and parameters can be
adjusted to match observations.

But I think communications traffic resulting from the events of 9-11
would have overloaded most systems so badly that we can not make
accurate measurements or predictions.  How many calls failed because
of trunk congestion?  Probably we don't know, because the number
probably exceeded the counting and storage capacity for such
information.  Probably the dial-tone capacity of many telephone
central offices was frequently exceeded, so we don't even know how
many calls of any sort were attempted.  Likewise for the radio
channel availability of wireless systems.  We don't know how many
people picked up wireless phones and could not even get a
control-channel assignment to begin to make a call.  Yet some people
in the planes and in the buildings were able to get through.  I
think the effect was initial shock, a vigil by the radio and TV news
sources, and then - a little later - a flood of calls.

We could probably put together a rough idea of the numbers.  But
even if we could evaluate the service demand in such emergencies and
attempt to meet it, I am sure that nobody would want to pay the
cost, so it won't happen.  This is why I have strongly opposed
proposals for the use of standard public communication facilities as
a primary link for emergency communications.  The best that could be
done would be to have an emergency service bit or field in the ID of
certain wireless phones.  Then systems could recognize and give
immediate service to emergency communications users.  I'm not sure
that would be practical, though.  Who is going to supervise this
across all of the phone systems and all those emergency phones?  And
wired phones... would emergency workers have to carry a sound-box
with them, and use it when calling from any wired phone where they
happen to be?  Even assuming an overloaded CO could even notice and
respond to an emergency signal at all.

We still need radio hams.  They can get it done, somehow, no matter
what.


----- Original Message -----
From: "Alex Fraser" <beatnic@comcast.net>
To: <tacos@amrad.org>
Sent: Friday, May 17, 2002 4:57 PM
Subject: Re: Erlang?


>     Well generally to start anyway. Seems it might be a good tool
to explain
> what happens in disaster communication situations.  I have not
seen any
> descriptions of 9-11 from a point of view of channel capacity, all
services
> included.
>     Would Erlangs now be considered part of information theory?