Cryptanalysis of AES-128 and AES-256 block ciphers using Lorenz information measure

Abstract EN

Encryption algorithms will transform a human interpretable text block or information in to a non-interpretable block of symbols.

The objective of any such encryption algorithm will be making the cipher block more non-interpretable and seemingly random block of symbols.

So any cipher block will always be random and will purely be a set of random permutations of symbols.

The efforts of distinguishing the cipher text of a cipher from random permutation and distinguishing a cipher blocks of different algorithms are called as “distinguisher attacks”.

Generally, almost all the classical ciphers are distinguishable and even breakable.

But the modern ciphers have been designed to withstand against several kinds of attacks and even withstand against distinguisher attack.

It means, we cannot even guess the type of cipher used for encryption only by seeing/analyzing the encrypted block of symbols.

In this work our focus will be only on distinguisher attack on modern ciphers.

For that, we have attempted to distinguish the cipher blocks of AES-128 and AES-256 using a metric called Lorenz Information Measure (LIM) which is commonly used in image and signal classification systems.

In our findings, we showed that the cipher blocks of AES-128 and AES-256 are certainly distinguishable from one another.