1993-07-09 - Hmm…hardware secure phone?

Header Data

From: eichin@cygnus.com (Mark Eichin)
To: tedwards@wam.umd.edu
Message Hash: 9955bef9d45d42fae74337346a70f6886454a48af762d77711cb05563c0b4d89
Message ID: <9307091919.AA05781@cygnus.com>
Reply To: <199307091816.AA09211@rac3.wam.umd.edu>
UTC Datetime: 1993-07-09 19:20:15 UTC
Raw Date: Fri, 9 Jul 93 12:20:15 PDT

Raw message

From: eichin@cygnus.com (Mark Eichin)
Date: Fri, 9 Jul 93 12:20:15 PDT
To: tedwards@wam.umd.edu
Subject: Hmm...hardware secure phone?
In-Reply-To: <199307091816.AA09211@rac3.wam.umd.edu>
Message-ID: <9307091919.AA05781@cygnus.com>
MIME-Version: 1.0
Content-Type: text/plain


>> Anyway, who knows of a DES or RSA chip which will do 16kbps?
>> Then all we need is a microcontroller to run the show, and 
	(kbits? kbytes?) Well, a quick check of some C code gets me
200Kbytes[*] (1.6Mbits) per second on a SparcStation ELC. I'll have
numbers for the i960 microcontroller[**] later this weekend -- you
might be able to have the microcontroller do the encryption too :-)
							_Mark_

[*] That's 50,000 calls to des_ecb_encrypt and 50,000 calls to
des_ecb_decrypt, so 100K ecbs, each ecb is 8 bytes, so that's 800K
bytes, and it took 3.98 user CPU seconds, so that's 200K bytes. Yep,
the math checks. Compiler: Solaris gcc -O (a few months old); DES
code: Ferguson's, as folded into Kerberos 4.

[**] Yeah, it's marketed as a microcontroller. If you put it in a
toaster, you wouldn't need a heating element. Still, it's part of the
background for my talk at the Embedded Systems conference in October
on "Security Issues in Embedded Networking". 




Thread