Path: utzoo!utgpu!jarvis.csri.toronto.edu!cs.utexas.edu!mailrus!accuvax.nwu.edu!nucsrl!telecom-request
From: nvuxr!deej@bellcore.bellcore.com (David Lewis)
Newsgroups: comp.dcom.telecom
Subject: Re: Modems and Phone Rates
Message-ID: <1892@accuvax.nwu.edu>
Date: 6 Dec 89 18:23:05 GMT
Sender: news@accuvax.nwu.edu
Organization: Bellcore, Livingston, NJ
Lines: 70
Approved: Telecom@eecs.nwu.edu
X-Submissions-To: telecom@eecs.nwu.edu
X-Administrivia-To: telecom-request@eecs.nwu.edu
X-Telecom-Digest: Volume 9, Issue 556, message 4 of 7


In article <1798@accuvax.nwu.edu>, jbayer@ispi.com (Jonathan Bayer) writes:

> david@elroy.jpl.nasa.gov (David Robinson) writes:
 
> > From the discussion so far it appears that modems do not take up
> >anymore phone network resources than a normal voice call, you get the
> >same ~4KHz bandwidth whether you are talking or using a modem.
 
> Sorry, you're wrong.  The telephone network is designed to work with
> human voices. As such the equipment multiplexes many conversations
> onto a single wire.

Yes, but the multiplexing method you describe is not the one used in
general.

>  Human conversation has many gaps that the network
> can use to multiplex other conversations using the same frequency.  A
> modem is on continously, tying up a frequency full-time.  Assuming
> that a wire can handle 100 different conversations at one time, and
> further assuming that 10 % of the conversations is quiet, that means
> that with the proper equipment a single wire could handle 110
> conversations at the same time. 
> I am sure that my numbers are not correct, but the method is valid.  

Valid, perhaps.  Used, no.  (OK, before someone jumps on me and starts
throwing "statistical TDM" around... not used by the public switched
telephone network in any major applications.)

> I do not support the idea of extra charges for modem usage, and the
> phone companies' numbers will have to be looked at very carefully,

OK, let's clarify some terms.  A two-way voice conversation includes
energy in the frequency band 20-20000 kHz.  The majority of this
energy is below 4000 Hz.  An intelligible voice conversation,
therefore, can be considered to include energy in the frequency band
300-3300 Hz.  It also includes a large amount of dead air.

A telephone voice channel is capable of carrying energy in a frequency
band from about 300 to about 3300 Hz.  This channel is constantly
available, end to end, to the user.  Regardless of the fact that no
energy may be carried at a given point in time, the capacity is
immediately, fully, directly available to the end user at any given
point in time, and is not used by the network for any other purpose.

This is true whether you're talking an analog loop, an analog trunk, a
time-division multiplexed digital trunk (on any medium), an ISDN
loop...  doesn't matter.  The 3000Hz of capacity is not used by the
network for any other purpose.

Therefore, it matters not whether that 3000Hz of capacity is 100%
utilized, 90% utilized, or 10% utilized -- the resources are fully
available to the end user, and you should not charge the end user more
because he's able to make more efficient use of the channel provided
him.

> however you cannot deny that modems _do_ take up bandwidth that
> conversations do not.

Yes, I can.  I can't deny that modems make more efficient use of the
available bandwidth -- but I certainly can deny that modems "take up
bandwidth conversations do not".

A modem, be it a 300bps Bell 103, a 1200 bps Bell 212a, or a 9600 bps
V.32, uses 3000 Hz of bandwidth.  Period.  A conversation uses 3000 Hz
of bandwidth.  Period.
 
David G Lewis					...!bellcore!nvuxr!deej
	(@ Bellcore Navesink Research & Engineering Center)
			"If this is paradise, I wish I had a lawnmower."