Consider a block cipher that encrypts bit strings of length $n$, where the key-space of the block cipher is of size $2^{kn}$, $k \geq 1$. My understanding of perfect secrecy is that a system is perfectly secret if (and only if?) the size of the key space is larger than the size of the message space. But then, if the stated block cipher was used to encrypt messages, is it not true that such a block cipher could encrypt $k$ messages of length $n$ with perfect secrecy? Would this be possible if, after each encrypted message, a nonce was used that changed the key in some way, so that every subsequent message had perfect secrecy?
-
$\begingroup$ Related Key length and message length in AES and perfect secrecy if plaintext length <= key length $\endgroup$ – kelalaka May 18 '20 at 10:14
-
$\begingroup$ And also Can one claim that AES is a perfect cipher? and Can one claim that AES has perfect secrecy for a key size and message size of 128 bits? $\endgroup$ – kelalaka May 18 '20 at 10:21
It is helpful to first define what a perfectly secure block cipher is. In the ideal case there are n different keys where each key uniquitely indexes one of the permutations mapping {0,1}^k to {0,1}^k. In this case we have n=(2^k)! Possible permutations/keys.
Perfect security typically means that no information is revealed at all. Not even against an adversary that is computationally unbounded (ie brute force) .
Given such an ideal block cipher (random permutation), can you think of a situation where information is leaked after seeing more than one message. Keep in mind, brute force attacks are allowed here.
Hint, this is why we have block cipher modes.