0
$\begingroup$

Let's suppose I want to convert a Substitution-permutation network block cipher like Kuznyechik to a version with a larger block size.

Should its S-boxes be enlarged too in order to achieve similar security?

$\endgroup$
4
  • $\begingroup$ This is quite a challenging problem. You might look at Rijndael Design which supports a 256-bit block size. $\endgroup$
    – kelalaka
    Commented Mar 24, 2022 at 21:36
  • 1
    $\begingroup$ @kelalaka I took a look in a implementation of Rijndael wich has a option to use 256-bit block size: github.com/charltoncr/Rijndael-Cipher -- It uses the same S-box for all the block sizes, I believe that this applies also Kuznyechik as it uses two 64-bits words (t1 and t2) which forms the processed block, each one is mixed with the S-boxes, I believe putting 4 64-bits words would be the same. Thanks. $\endgroup$ Commented Mar 25, 2022 at 1:32
  • 2
    $\begingroup$ Be careful, in SPN network, (the Permutation is really important) make sure that the permutation really works as intended. The RijnDael team was carefully calculated the necessary rounds. You may look at their book about this, too. $\endgroup$
    – kelalaka
    Commented Mar 25, 2022 at 9:15
  • $\begingroup$ @kelalaka I got. $\endgroup$ Commented Mar 25, 2022 at 19:28

1 Answer 1

1
$\begingroup$

Usually not, but increasing block size is cipher dependant.

Increasing block size usually involves change to linear part of a cipher and probably increasing number of rounds.

For example you could use bigger MDS (like Whirlpool), use different permutation (like Rijndael), or add something additional like Pseudo-Hadamard transform (like Twofish).

Another way would be to use round function as F in Feistel.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.