From: Andrew Loewenstern <andrew_loewenstern@il.us.swissbank.com>
Date: Tue, 12 Sep 95 10:03:01 PDT
To: Mike Rosing <cryptech@Mars.mcs.com>
Subject: Re: Elliptic Curve Public Key Crypto available
Message-ID: <9509121658.AA00550@ch1d157nwk>
MIME-Version: 1.0
Content-Type: text/plain


>  Previous to the availabilty of 100 MHz processors this method of
>  public key crypto was ridculously slow.  The only versions were
>  laboratory curiosities implemented in ASIC hardware.  Code, if it
>  exists, is not in the public domain.

What about NeXT's Fast Elliptic Encryption (FEE)?  Their research guy,  
Richard Crandall, came up with major speedups to elliptic curve encryption.   
In fact, there was a simple cryptosystem that shipped as a demo with NeXTSTEP  
2.0.  What was most interesting about this system was that it didn't store  
any keys anywhere;  your public/private key pair was generated _on the fly_  
from your passphrase every time you encrypted or decrypted....on a 25mhz  
68040 too...   and it was fast!  It wasn't that great of an implementation  
(you _really_ need a lot of bits of entropy in that passphrase, and you can't  
change your passphrase without changing your PK), but it shows how fast  
NeXT's speedups are.  And this was in 1990...

I'm not sure if the speedups are patented, but you could try a literature  
search.  If it really is fast then it could mean good things for servers that  
need to do a lot of enrcyption/decrption for interaction with clients.


andrew