Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Posting-Version: version B 2.10.2 9/18/84; site mouton.UUCP
Path: utzoo!watmath!clyde!burl!ulysses!allegra!mouton!karn
From: karn@mouton.UUCP
Newsgroups: net.ham-radio
Subject: Re: packet help
Message-ID: <220@mouton.UUCP>
Date: Fri, 9-Nov-84 01:43:04 EST
Article-I.D.: mouton.220
Posted: Fri Nov  9 01:43:04 1984
Date-Received: Sat, 10-Nov-84 07:42:55 EST
References: <168@zps.UUCP> <13200011@smu.UUCP>
Organization: Bell Communications Research, Inc
Lines: 51

Regarding the suggestion to use voice repeaters for packet, I have some
comments.

1. Voice repeaters are usually not designed for fast, bursty transmissions
such as those produced by packet boards. Keyup delays can involve quite
a bit of overhead, particularly if mechanical relays are involved. The
constant "kerchunking" involved in packet can cause quite a bit of wear
and tear on these relay-switched machines (and on their tech-committee
chairmen.)

2. Voice and packet users simply cannot coexist peaceably on the same
repeater. The "racket radio" users drive the human listeners to distraction
(mainly because they can't understand what's being said about them!), and
the long-winded voice users can bring data traffic to a standstill.

3. Because of carrier hang timers on most voice repeaters, the TNCs cannot
use RF carrier detect for CSMA (not that many people do, even on simplex),
but must use modem subcarrier detect. This also has the effect of infuriating
the voice users, since they get stepped on by the packet boxes.

4. A voice repeater is very expensive in comparison to a single
frequency "digipeater", since it requires cavities and isolation to
run in full duplex.

5. The one advantage of using a voice repeater over a single frequency
digipeater is that collisions due to one station's inability to hear all
others on the channel can be avoided. However, working against this is
the fact that a wide-coverage repeater forgoes the possibility of "frequency
reuse", in which two pairs of stations who are far enough apart could
communicate simultaneously on the same frequency without interference,
given enough of a signal-to-interference ratio to get good capture.

There is an opportunity for a research project here that would be of real use
to amateur packet radio. Do an analysis to find, given an actual set
of user stations and antenna QTHs, the optimal power level for each station
to make maximum utilization of a channel, considering the effects of
digipeating, collisions, etc, and to compare the results to the efficiency
of a wide-area duplex analog repeater. One might start by looking at the
work of Kleinrock and others and seeing how it might be applied to a
practical situation.  I remember one paper of his in which he showed that
a packet radio network did not have to be "disconnected" (i.e., have
pairs of stations that could not directly communicate) to a very great
degree before the channel throughput degenerated to the Aloha level.
It would be interesting to see how this bears out in an actual situation.

In any event, my practical advice is to put up a single-channel digipeater
in a good location if you want to provide a utility service to the local
packet radio community.

73, Phil, KA9Q