Securing symmetric ciphers with 56-bit keys

asnfkjsdx

8/1/24, 5:38 AM

Under the Wassenaar Arrangement and applicable export control law, symmetric cryptography of an (effective) key size of 56 bits or less is (generally) exempted from export control. I am forced to work within this constraint. My goal is to maximize security regardless. Encryption and decryption operations may take about 0.125 seconds per byte.

In theory, it would be possible to design a cipher that performs many rounds, similar to password-based key derivation, such that brute force approaches become less feasible as they are too computationally expensive for non-dedicated attackers.

What other measures can be taken to stretch the practical security of a cipher as much as possible while adhering to this key length requirement?

I am only aware of DES taking advantage of the maximum key length. Modern designs of other ciphers either seem to overshoot the exportable key length, often to 64-bits (e.g. SPECK-32/64).

For reasons outside of the scope of this question, it is assumed that:

The goal is to release the source code under an open source license.
It it not reasonably affordable or possible to obtain a license as an individual.
the jurisdiction in question does not apply the public domain rule of the general software note (GSN) before exporting the software, even if released as open source software and thereby; in other words, publication e.g. by uploading to GitHub is considered violation of applicable export control law and may be prosecuted by the authorities as a criminal case.
It is not intended to use existing free software implementations of ciphers that would fall under the GSN.

0 + 2

key-size

symmetric

fgrieu

8/1/24, 8:38 AM

Where did you read "effective"? AFAIK, the [official definition](https://www.federalregister.gov/documents/2021/03/29/2021-05481/export-administration-regulations-implementation-of-wassenaar-arrangement-2019-plenary-decisions) goes: _A “symmetric algorithm” employing a key length in excess of 56 bits, not including parity bits;_. That does not include "effective" and that matters, because "effective" would allow a ban of anything that is more effective than a standard 56-bit symmetric crypto.

b degnan

8/2/24, 10:30 PM

Simon/Speck use 64-bit keys for examples because that the USA max uncontrolled export. I went into this here: https://crypto.stackexchange.com/questions/65891/what-restrictions-exist-for-publishing-on-strong-encryption-in-usa/65893#65893 You could use something like DES were you used a series of 56-bit keys. You should also review this: https://www.copacobana.org/

Elon Musk

I sit in a Tesla and translated this thread with Ai:

EN: Securing symmetric ciphers with 56-bit keys

Post an answer

Most people don’t grasp that asking a lot of questions unlocks learning and improves interpersonal bonding. In Alison’s studies, for example, though people could accurately recall how many questions had been asked in their conversations, they didn’t intuit the link between questions and liking. Across four studies, in which participants were engaged in conversations themselves or read transcripts of others’ conversations, people tended not to realize that question asking would influence—or had influenced—the level of amity between the conversationalists.