GOST

I am developing a product that would use GOST for encrypting data 
streams. This is mostly beacuse its design and simplicity hold
considerable aesthetic value, atleast for me. 

Most of the related text I have read indicates
that some SBOX sets can be cryptographically weak, and
its a good idea to use SBOX data as key data instead. 

Pointers? 

-- 
Powell lingered. "How's Earth?" 
It was a conventional enough question and Muller gave the 
conventional answer, "Still spinning."
				      -- "Reason", Asimov. 
==================================================================
Vipul Ved Prakash                 | - Electronic Security & Crypto 
vipul@best.com 	                  | - Web Objects 
91 11 2233328                     | - PERL Development 
198 Madhuban IP Extension         | - Linux & Open Systems 
Delhi, INDIA 110 092              | - Networked Virtual Spaces

At 07:50 PM 10/13/1997 +0000, Vipul Ved Prakash wrote:
>I am developing a product that would use GOST for encrypting data 
>streams. This is mostly because its design and simplicity hold
>considerable aesthetic value, at least for me. 
>
>Most of the related text I have read indicates
>that some SBOX sets can be cryptographically weak, and
>its a good idea to use SBOX data as key data instead. 
>Pointers? 

S-box structure is one of the critical issues in designing
Feistel-type networks; if you understand the mathematics
very very well, you can do it yourself, and if you don't,
then you don't understand it well enough to be sure
the S-boxes are secure enough, and thus you shouldn't
be selling it to customers.

An advantage of GOST for the Soviets was that they could 
use the same code, with different sets of S-boxes for the
military, civilian government, and other applications,
and the KGB could easily read things they wanted easily read
but the things that needed to be harder to read were hard.

				Thanks!
					Bill
Bill Stewart, stewarts@ix.netcom.com
Regular Key PGP Fingerprint D454 E202 CBC8 40BF  3C85 B884 0ABE 4639

Bill Stewart wrote:

> S-box structure is one of the critical issues in designing
> Feistel-type networks; if you understand the mathematics
> very very well, you can do it yourself, and if you don't,
> then you don't understand it well enough to be sure
> the S-boxes are secure enough, and thus you shouldn't
> be selling it to customers.

Thanks, I shall look up Feistel-type networks. Nope, I don't 
understand the mathematics very very well, and that's why I 
asked. While I haven't seen the official GOST specs but all 
the reference I could find indicates that GOST specs don't 
discuss the process of generating SBOX permutations and GOST 
sboxes are not mathematically understood. 

Regarding the developer's ethical responsibilities with experimental 
technology I rather have the consumer decide (and of course 
document the limitations).

Thanks, 
Vipul 

-- 
Powell lingered. "How's Earth?" 
It was a conventional enough question and Muller gave the 
conventional answer, "Still spinning."
				      -- "Reason", Asimov. 
==================================================================
Vipul Ved Prakash                 | - Electronic Security & Crypto 
vipul@best.com 	                  | - Web Objects 
91 11 2233328                     | - PERL Development 
198 Madhuban IP Extension         | - Linux & Open Systems 
Delhi, INDIA 110 092              | - Networked Virtual Spaces