Ciphertexts Clustering is Equivalent to Plaintexts Clustering

Abstract

Several studies have been made in attempt to break confidentiality, either by obtaining the knowledge of the plaintext or the key itself working only with cryptograms. However, there is not known methods capable of breaking contemporary cryptographic algorithms, as DES and AES. Nevertheless, in order to benefit cryptanalysts, it is possible to search weakness in these algorithms. In this work we show that ciphertexts can be considered as plaintexts written in an unknown idiom and using a binary alphabet, where each idiom is determined by the cryptographic key. In the experiments with ciphertexts and plaintexts clustering it have reached success, since all ciphertexts encrypted with the same key belong to the same group, as well as, plaintexts, written in the same idiom and alphabet belong to the same group. This result exposes a cryptographic algorithms weakness, since they are designed to generate ciphertexts without any relation with the input data, such as the plaintext or the cryptographic key.