From: jim bell <jimbell@pacifier.com>
Date: Mon, 19 Feb 1996 08:35:42 +0800
To: SINCLAIR  DOUGLAS N <dsmith@midwest.net (David E. Smith)
Subject: Re: Using lasers to communicate
Message-ID: <m0toIW1-0008zJC@pacifier.com>
MIME-Version: 1.0
Content-Type: text/plain


-----BEGIN PGP SIGNED MESSAGE-----

At 10:44 PM 2/17/96 -0500, SINCLAIR  DOUGLAS N wrote:
>> If you have a secure link you don't need encryption.  Arguably, the
>> converse is true; if you have secure encryption you don't need
>> a secure link.  Isn't the ability to transmit secure data over
>> insecure channels one of the primary justifications for encryption?
>> 
>
>Of course.  My point, though I seem to have failed to state it,
>is that encryption is a cheap software thing while laser beams
>are expensive, complicated, and still not secure.

I tend to agree.  But my position is a bit more "middle of the road":  We 
_should_ use laser/LED links, but we should encrypt the link with encryption 
sufficiently strong (IDEA/1024-bit RSA key) to make interception of the beam 
pointless.

I think what's needed from the IC companies is a chip somewhat analogous to 
the UARTS (TR1602/AY5-1013) (which were "new" in about the 1975 time frame), 
but one which maintains one half of a bidirectional link with NSA-proof 
encryption.  It wouldn't matter what the physical medium was, it would 
"handle it."  They'd be given "authority" over link signal amplitude, and 
would be able to monitor link integrity/error rate to anticipate incipient 
link failures.  (caused by electronic/mechanical  failure, growth of 
vegetation, corrosion, and other items.)  (I know, I know, shades of 2001!  
"Open the pod bay door, Hal!")

During periods of low usage, it would occasionally automatically engage in 
link margin testing, etc, and automatically generate/transmit extremely-long 
period pseudorandom data to  prevent snoopers from doing any sort of
traffic-density analysis 
on the working link.  If the chip was given mechanical authority over 
beam-pointing, the chip could also do auto-align test functions to
compensate for 
misalignment, etc.  Alignment would be kept "perfect."

While I'm no IC-design expert, considering the fact that chips commonly 
possess at least 1000 times as many transistors as they did in 1974, they 
SHOULD be able to implement such a chip easily enough.

Lazy bastards.

Jim Bell
jimbell@pacifier.com



-----BEGIN PGP SIGNATURE-----
Version: 2.6.2

iQCVAwUBMSezPPqHVDBboB2dAQGu1gP+J1t3aagiHVoSE0ETiid2TPYw1wCBxi5H
znvWIHbic9VNMnBo1ZkeSiR86Xi/C311CB526vRZnzNyUNuk8vF55MxGY7FRf1sn
xGvH0n+b3Y4XR3NsJP0cazLhmDZocjTTjiRDGMSFt4wwLt0SqiLbrxQ/WkcB6ee+
/17ORpzAafk=
=klc5
-----END PGP SIGNATURE-----