Looking for fast KASUMI implementation

[Jack Lloyd]

Define fast - KASUMI is based heavily on MISTY1. In fact, during a fast scan of
the KASUMI spec, I couldn't see anywhere obvious where it different from MISTY1
at all. As far as I know, I'm the only person who has even tried writing fast
code for MISTY1, and the result is quite dog-slow compared to most other common
ciphers (to pull some numbers out of the air: around 4.3 MB/sec on an 800 MHz
Athlon, compared with 9.4 MB/sec from AES-128 and 15 MB/sec from 16-round
RC5). Obviously you can do better on a faster processor (and I'm sure there are
some cycles yet to be squeezed out of my MISTY1 code - there are many who can
hand-optimize better than I), but I don't think MISTY1 (or KASUMI) will ever be
very fast in software.

Would a FPGA work instead? That seems like your best bet to me.

-Jack

On Thu, Dec 15, 2005 at 08:24:23AM -0500, james hughes wrote:
> Hello list:
> 
> I have research project that is looking for a fast -software-  
> implementation of the KASUMI block cipher.  I have found many papers  
> on doing this in hardware, but nothing in software. While free is  
> better (as is beer), I will consider a purchase.
> 
> FYI, KASUMI is the cryptographic engine of the 3GPP.
> 	http://en.wikipedia.org/wiki/3gpp
> 
> Thanks.
> jim
> 
> 
> ---------------------------------------------------------------------
> The Cryptography Mailing List
> Unsubscribe by sending "unsubscribe cryptography" to majordomo at metzdowd.com

---------------------------------------------------------------------
The Cryptography Mailing List
Unsubscribe by sending "unsubscribe cryptography" to majordomo at metzdowd.com