21 ms·
Post-quantum crypto proposal using Transcendentals
Transcendental numbers expand as a series of random, non-repeating digits (think Pi).
Assume Alice could find a new transcendental number between 0 and 1. (There are an infinite number of them).
Assume Alice and Bob know that number. This is their shared, private key. PRIVATE
To communicate, Alice creates a new transcendental number using the private key (for example by exponentiation, or some other offset method). This "offset" OFFSET is the public key.
To encode, Alice starts with the private key, creates a new transcendental number using some random offset. Call the new transcendental number PAD = f(PRIVATE, OFFSET). Alice uses PAD as a one-time pad to encrypt the message SECRET = (MESSAGE xor PAD)
Alice sends the pair (OFFSET,SECRET) to Bob.
Bob uses OFFSET to compute the PAD from PRIVATE. Then uses (SECRET xor PAD) to recover MESSAGE.
Transcendental numbers plus offset can generate an infinite set of one-time PADs.
Knowing (MESSAGE xor SECRET) recovers the PAD so this in insecure against a known plaintext attack. Knowing PAD then PRIVATE can be recovered. However, if Alice and Bob share a secret method of choosing a new PRIVATE after each communication then they can continue the scheme.
The key question is how to find transcendental numbers.
- HelloNurse 5y ago> The key question is how to find transcendental numbers. That's indeed difficult and ad-hoc, particularly considering that you'd need to derive them from the shared private key rather than finding any one in a large family. Typically, pseudo-OTP schemes use a plain old pseudorandom number generator that "inflates" a large but finite secret key to an approximationof a true OTP. Using transcendental numbers for that is complicated and inefficient; what advantage over conventional technology do you expect?