2
$\begingroup$

I did found an old question which answers exactly this but it is from 2 years ago and i was wondering if there were any changes. I have not been able to find any new information on the cipher and am quite curious seeing as it's extremely fast.

I've tested it against a very fast (fastest I have found so far) AES implementation which took 392μs to encrypt 128 bytes and 400μs to decrypt 128 bytes.

The SEED cipher took 325μs to encrypt the same 128 bytes and 228μs to decrypt. This is enough to make it interesting for me though I still prefer AES since it is more well known and has been through more scrutiny.

Both tests were done with the same CBC method, so there might be some performance to be gained from a better CBC implementation but i think the one i have is already quite good, for those of you who are wondering i'm using the CycloneSSL library on a RX63N board (renesas development board).

So can SEED be trusted? Can i use it for sensitive data?

$\endgroup$
  • $\begingroup$ In my experiences SEED is noticeably slower than AES on x86_64. $\endgroup$ – LightBit Mar 24 '15 at 19:22
  • $\begingroup$ I'm also wondering, why your SEED is much faster when decrypting? $\endgroup$ – LightBit Mar 24 '15 at 19:32
  • 3
    $\begingroup$ You can't really test with just so little plaintext. The overhead is much too High. Try a few megabytes. $\endgroup$ – Nova Mar 25 '15 at 8:58
  • 1
    $\begingroup$ @VincentAdvocaat You can encrypt same block multiple times (10000x). If you need fast stream cipher I would suggest ChaCha. $\endgroup$ – LightBit Mar 27 '15 at 8:36
  • 1
    $\begingroup$ If you are on an embedded device, you should consider using an implementation secured against side-channel attacks. When the AES was chosen, one criterion was how difficult it is to implement countermeasures against side-channel attacks. SEED is a bit more problematic (it needs switching from additive to boolean masks). $\endgroup$ – j.p. Apr 8 '15 at 13:57
1
$\begingroup$

Short answer: Yes, I'd still trust SEED.

The topic you linked to is from 2013 (2 years) old and for cryptography nearly never any advances will be made for a 15 year old, rarely used cipher after just 2 years.

As the best cryptanalysis only penetrated half of the rounds back then I'd highly doubt there're any advances by now, as this is only a legacy cipher in most systems and hence not a target like AES which is in widespread use.

EDITED TO ADD(1): As pointed correctly in a comment, if you plan to secure data against nations / national intelligence agencies, you can not rely on the security of SEED and therefore must consider using AES(-192/-256) or maybe even multi-encryption with ciphers with different designs (-> SEED & AES & Threefish/ChaCha & Serpent).

$\endgroup$
  • $\begingroup$ Not all analysis is public. Just to see no advantages in the field doesn't mean that the algorithm is still secure. So by itself this doesn't really answer the question. $\endgroup$ – Maarten Bodewes Apr 6 '15 at 17:08
  • $\begingroup$ the only ones not publishing their analysis are intelligence agencies and such if they are your advesaries, you need something more new, like AES/Serpent/ChaCha/Twofish/Threefish. $\endgroup$ – SEJPM Apr 6 '15 at 17:15
  • $\begingroup$ I can live with that conclusion :) Especially because SEED is a pretty conservative design (in the sense that it doesn't do anything out of the ordinary). $\endgroup$ – Maarten Bodewes Apr 6 '15 at 17:19

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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