Path: utzoo!utgpu!watmath!clyde!att!osu-cis!tut.cis.ohio-state.edu!cwjcc!neoucom!wtm
From: wtm@neoucom.UUCP (Bill Mayhew)
Newsgroups: comp.dcom.modems
Subject: Re: V32 will smash TB+ in a year (?)
Summary: echo cancelation and satelites
Message-ID: <1406@neoucom.UUCP>
Date: 13 Nov 88 15:57:11 GMT
References: <10805@cup.portal.com> <1417@percival.UUCP> <11078@cup.portal.com>
Distribution: na
Organization: Northeastern Ohio Universities College of Medicine
Lines: 53


One subject area that has not been discussed very much is the use
of modems in conjunction with satelite communications.  Only a few
of the v.32 modems that I have tried have worked well with very
long echo delays (150+ mS).  Most notably, AT&T's v.32 modem seemed
to be able to handle anything we gave it.  I experienced some bugs
with the UDS modem.  I only had one Concord modem at the time, and
had a lot of trouble getting it to establish a connection with
other brand v.32 modems.  Interestingly, price versus performance,
seems to relate to v.32 modems too; the AT&T is one of the most
expensive offerings, but it works the best too.

There are places in the world were it might be difficult to avoid a
multihop satelite link.  The Indian Ocean area comes to mind; there
are relatively fewer undersea cable links in that area.  Since
satelites are in an orbit roughly 36,000 km obove the earth's
surface, the round trip delay going through a satelite is about 240
mS.  If one happened to get a double hop, that is 480 mS of delay.
That is probably outside the margin of virtually any v.32 modem to
deal with.  One might not see a double hop in the direct dial
network, but if one's company has a large private network of leased
transponders, the odds are worse (ie, favor double hop).

In cases such as the above, you will win if you pick an asymetrical
protocol and PEP modem modulation.  Unix's g protocol together with
a trailblazer or ventel pathfinder PEP modem makes a nice combination.
PEP modems will also be relatively less impacted in throughput on a
poor link to such areas too.

In the domestic US and european regions, v.32 modems make more
sense with a duplex protocol such as FAST or whatever.  There are
many terrestrial fiber optic links available, so the odds of going
through a satelite are smaller.

Users such as Rick Adams at seismo have reported quite good performance
data with PEP modem links to far flung locations with long delays or
noisy links.

As far as I am concerned, not having both a PEP modem AND a v.32
is a good business practice.  You can trust me that management
doesn't give a damn what you buy as long as they see the operating
costs of your department decreasing.  Capital goods such as modems
can be depreciated, allowing the costs to be recovered over time.
Operating expese for telecommunications makes management angry
because that is money that is gone.  If I have anything that lowers
my expense, it makes me look good.  PEP modems and V.32 modems both
make me look good.  For the moment, most of our communications is
via the direct dial network, but in the near future that will
switch over to a leased T1 circuit with a 56Kbaud interface that is
neither PEP nor V.32, but we'll probably spend less money that way
and management will be happy.

--Bill