From: jim bell <jimbell@pacifier.com>
Date: Fri, 5 Jan 1996 15:03:49 +0800
To: frantz@netcom.com (Bill Frantz)
Subject: Re: Guerilla Internet Service Providers (fwd)
Message-ID: <m0tY4b4-00092hC@pacifier.com>
MIME-Version: 1.0
Content-Type: text/plain


At 10:54 AM 1/4/96 -0800, you wrote:
>At 11:43 1/4/96 -0600, Jason Rentz wrote:
>>Previous exchanges deleted...
>>>
>>>With a tightly focused beam (light is easy, I don't know about lower
>>>frequencies), you can prevent interception except by very obvious physical
>>>devices.  (e.g. Someone in a cherry picker truck.)  You may be able to
>>>avoid the need to encrypt the link (and all the paranoia about key
>>>management, advances in factoring etc. that that implies.)
>>>
>>>Bill
>>
>>The problem with this comes when you start creating links between much
>>taller buildings like in San Fran.  Any give building over 30 stories might
>>sway a foot or so at any given time.  Combine that with the other building
>>and you might get a few feet of movement. (movement not including during an
>>earthquake)  :)
>
>(1) No single communication technology is appropriate for every problem.
>
>(2) A technical fix could include having the receiver send steering orders
>to the transmitter.  This solution would, of course, be a long way from the
>low tech scavenged lens and 1/2 meter cardboard mailing tube technology I
>was thinking of.

I think you guys (further up the reply chain) are missing the point.  While
IR does have stealth advantages in, say, wartime, for routine network usage
everyone can be assumed to know where everyone else is, and where all the
optical links are, etc.  There's no point trying to use link-location
secrecy.  And presumably, encryption will provide all the
message-secrecy/anti-spoofing functions required.  Simply ASSUME that the
beams can be intercepted (although probably not intentionally cut).  That's
why we're "cypherpunks," right?!?

Secondly, IR beams can be plenty narrow enough to avoid inter-link
interference, but at the same time wide enough to avoid beam-steering
problems. Note: I'm assuming link distances of under, say 300 meters here.

Previously, a point was made about the effects of fog cutting links:  Due to
scattering, one of the reasons automobile fog lamps are 550 nanometer
yellow/orange is to minimize the scattering that shorter wavelengths (400 nm
blue, 450 nm green) are more prone to.  I would imagine that near IR at,
say, 890 nm would be dramatically less sensitive to such scattering.  1400
nm might be even better.  Rain might be a different story.  But then again,
if we're limiting the links to around 300 meters, the total amount of water
between "here" and "there" CAN'T be all that great.  And in addition, one of
the advantages of computer networking over telephone-type networking is that
we can "tolerate" (although, not LIKE) the occasional necessity of
re-transmitting data.  And dynamic re-routing is probably far easier than
for real-time telephone-type data.

>From the standpoint of computer networking, the main benefit of IR is to
cross rights-of-way without permission or trenching (or stringing cables
from telephone poles) in urban and suburban areas, allowing data transfer
near-fiber speeds.   In an urban setting, a single tall building could
become a central hub for most of its nearest neighbors.   I don't anticipate
IR being used "to the home" (especially since residential areas have trees,
etc); rather, I would imagine that it would be used to feed the occasional
top-of-the-telephone-pole microcell, with very-low-milliwatt (or high
microwatt) RF going the last 100 meters or so to the home.  This would allow
a non-phoneco, non-cableco company to offer bidirectional networking in an
entire residential area with an absolute minimum of costs/rights aquisition.