-3
$\begingroup$

Proposed Cipher suite (using aes-ecb):

cipherText = aes-ctr(key, ++iv, (plainText)); & authTag= aes-ctr(key, iv, aes-ecb(key, sha-1(cipherText+authData+key+iv)));

Security targets:

  1. Encryption of the plainText.
  2. Integrity of the cipherText & authData.
  3. Authenticity of the cipherText & authData.

Used components and reason behind using them:

  1. sha-1: For checking integrity of the cipherText & authData.
  2. aes-ctr: Used for encryption of the plaintext & aes-ecb(sha1).
  3. aes-ctr(aes-ecb(sha-1(***cipherText+authData+key+iv)))*:
    • sha-1 is encrypted with aes-ecb to create hurdle in finding out the key stream used by aes-ctr to encrypt `aes-ecb(sha1). Because even if the attacker get to know sha-1, sha-1's encrypted value could not be guessed.
    • sha-1 is used to encrypt the cipherText & authData along with key + iv: So that we can check integrity of cipherText & authData. key + iv is added to create hurdle for attackers to guess value of sha-1.

The challenge/question is to find out security level of this cipher suite.

This question is for passionate & elite cryptanalysts, which would help them or other (from their answers), to understand how use of a broken hash function determines overall security?

I hope that some people would find this as a interesting challenge.


*Please note that value of sha-1 is truncated to block size.(i.e, 128 bits)

This method is improved version derived from the method proposed in this question.

$\endgroup$
3
$\begingroup$

I was just trying to build a secure enough cipher suite using aes and a sha-1 function

Your safest bet for this is to use AES-CTR and apply HMAC-SHA1 on the ciphertext. This should still be secure (because HMAC doesn't rely on collision resistance and mitigates length extensions). I.e. $$(c,\tau)=((\text{IV},\operatorname{AES-CTR}_{k,\text{IV}}(m)),\operatorname{HMAC-SHA1}_k(c))$$

Proposed Cipher suite (using aes-ecb):

I'm reasonably confident that this hits at least around a 64-bit security level, but the construction is too complex and too non-standard for me to trust my ad-hoc evaluation. Of course 64-bit is much less than you can get from more standard modes like GCM, CCM and EAX which should be preferred.


Additional nodes on sub-constructions:

Final conclusion:
Do not use this for applications where security actually matters!
When it doesn't matter and attackers will only invest a low amount of effort this has a chance of providing adequate security.

$\endgroup$
  • $\begingroup$ Well answered. This answer satisfies the need. But I am looking for an answer which can removes the "may be (partially)", and "ad-hoc evaluation" of this answer. Again thank you very much. $\endgroup$ – NKaran Dec 14 '18 at 18:00

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