In defining the notion of perfect secrecy, we first consider the threat model of a single ciphertext-only attack.
Adversary knows , and the probability distribution over .
Adversary passively eavesdrops on the communication and is able to get a single ciphertext.
No assumption is made on the computational power of the adversary.
Adversary does not know the secret key shared by Alice and Bob.
Definition. An encryption scheme with message space is perfectly secret if for every probability distribution over , every message and every ciphertext for which , it holds that
Lemma. An encryption scheme with message space is perfectly secret if and only if
or equivalently if
holds for every and every .
The adversary outputs a pair of messages .
A key is generated using and a bit is chosen uniformly at random.
The challenge ciphertext is computed and given to .
outputs a guess bit .
The adversary succeeds and the output of the experiment is defined to be if and otherwise. We write if the output of the experiment is .
Definition. An encryption scheme with message space is perfectly indistinguishable if for every it holds that
Lemma. An encryption scheme is perfectly secret if and only if it is perfectly indistinguishable.
Correctness: The One-Time Pad satisfies the correctness criterion, since for every we have
— Feb 1, 2023
Made with ❤ at Earth.