2,331 reputation
220
bio website crypto.stackexchange.com/…
location United States
age
visits member for 1 year, 8 months
seen Apr 11 at 21:24


Sep
30
answered Is it possible to create an open-source SecurID?
Sep
30
revised What are requirements for this grille?
Renamed from Cardan grille (incorrect) to Fleissner grille (correct)
Sep
30
revised What are requirements for this grille?
Added recommendation to edit question
Sep
30
answered What are requirements for this grille?
Sep
30
comment What are requirements for this grille?
I remember long ago reading some old books on historical ciphers talking about flipping and/or inverting the Cardano grille as an additional (rudimentary) key step. See en.wikipedia.org/wiki/Cardan_grille, which illustrates "Four positions of the grille", and an entire section on turning grilles here: en.wikipedia.org/wiki/Grille_%28cryptography%29#Turning_grilles
Sep
30
comment Are NIST's changes to Keccak/SHA-3 problematic?
80 bits is sufficient security for a symmetric encryption key; but I don't know if the same is true for a hash algorithm, where the attacks are very different.
Sep
25
comment Is there a developer's resource that ties encryption methods and strengths to the appropriate usage scenarios?
I think it's a fair question, when you look at it this way: what is a good approach to selecting cryptographic solutions? @RichieFrame 's answer below is a good one: start with a trustworthy document.
Sep
17
comment probability of N hash digits colliding
@kylek, the answer is still correct, regardless of what data you are hashing to get the five digits.
Aug
26
answered Is there a public key encryption scheme with optimal key size?
Aug
26
comment Multiple parties must encrypt and get the same result
@NickP, there are a few more details to consider. Make sure the GetToken(cardNo) service is as securely controlled as the GetCardNo(token) service, in order to prevent this attack. Also, my earlier statement about using sequence numbers as tokens could leak information about "unseen" vs "previously seen" numbers, so I wouldn't use sequence numbers as tokens.
Aug
23
comment Multiple parties must encrypt and get the same result
@NickP, the other concern is to view this from an attacker's viewpoint. He doesn't care if he gets 100% perfect results. If he can steal only 10% of your account numbers, that could be millions. Certainly enough to steal from.
Aug
22
comment Multiple parties must encrypt and get the same result
I think you're trying to solve the problem inside the oracle. But because it's a trusted and secure environment, he can use AES, SHA, FPE, it doesn't matter. What matters greatly is what is exposed to the users. If they are given a key and algorithm, as he asked for, it's game over. If they are given credentials and a service, he retains auditable control, which is critical for PCI compliance.
Aug
22
comment Multiple parties must encrypt and get the same result
Even bulking up the attack using scrypt or other make-work algorithm isn't enough to protect it. There are only 100,000 possible values that need to be tested if he's truly preserving 8 bytes of "non-personal details". It's just not a hard problem to brute force.
Aug
22
comment Multiple parties must encrypt and get the same result
Yes. To be secure, you can't distribute the encryption key or hash method (like he asked.) The cryptographic operation has to remain in a tightly controlled environment, like an oracle, and all access must be granted through a service that authenticates and authorizes its users. It still can't prevent the attack, but you could possibly detect one or trace one.
Aug
22
comment Multiple parties must encrypt and get the same result
No, it answers a question about encryption, but doesn't solve the security problem of the possible account number space being easily searchable.
Aug
22
comment Multiple parties must encrypt and get the same result
FPE is good if you're trying to fit 16 encrypted bytes in the place of 16 cleartext bytes. But that's not the problem they're trying to solve. They're trying to produce identical encrypted values for each account number, as he stated that multiple teams need to "arrive at the same result", so they could correlate or index encrypted account numbers. That requirement precludes random IVs or salts that would prevent such an attack.
Aug
22
revised Multiple parties must encrypt and get the same result
corrections
Aug
22
answered Multiple parties must encrypt and get the same result
Aug
22
awarded  Enlightened
Aug
22
awarded  Nice Answer