ENIGMA - JOURNAL OF INFORMATION SECURITY AND CRYPTOGRAPHY, VOL. 9, NO. 1, 2022 1  Abstract— This research presents a study on the identification of post-quantum cryptography algorithms through machine learning techniques. Plain text files were encoded by four post- quantum algorithms, participating in NIST's post-quantum cryptography standardization contest, in ECB mode. The resulting cryptograms were submitted to the NIST Statistical Test Suite to enable the creation of metadata files. These files provide information for six data mining algorithms to identify the cryptographic algorithm used for encryption. Identification performance was evaluated in samples of different sizes. The successful identification of each machine learning algorithm is higher than a probabilistic bid, with hit rates ranging between 73 and 100%. Keywords— Identification of cryptographic algorithm; Data mining; machine learning; post-quantum cryptography, NIST randomness tests. I. INTRODUCTION ryptology can be divided into cryptography and cryptanalysis. Cryptography can be defined as the science of encoded writing, ensuring that only the sender and recipient of a message have access to its content, thus providing confidentiality, irreversibility, authenticity and integrity of information [1]. Cryptanalysis is the science that aims to extract the plaintext from a ciphertext, without prior knowledge of the encryption key used [2]. To achieve its goal, cryptanalysis makes use of different types of attacks, and due to this characteristic it can be used to access the security of a cryptographic algorithm, making it essential for the development of modern cryptography [3]. C In a cryptanalytic scenario, little information is available besides ciphertexts and, a priori, it is not known which algorithm was used to encrypt the plaintexts. Therefore, the process of identifying the algorithm used in the encryption process considerably reduces the cryptanalysis effort and is part of the set of activities that contribute to the decoding of the message, which also includes the determination of the key size and the key itself [4]. The development of modern cryptographic algorithms is based on complex mathematical models, which aim to dissipate any patterns that may exist in the ciphertexts produced by them [5] and make difficult the process of determining the algorithm used. In the literature, there are several studies that analyze the task of determining cryptographic algorithms based on the  B.S. Rocha, Military Institute of Engineering (IME), Rio de Janeiro, Brazil, rocha.bruno@ime.eb.br J. A. M. Xexéo, Military Institute of Engineering (IME), Rio de Janeiro, Brazil, xexeo@ime.eb.br R. H. Torres, University of Pará, renatohidaka@ufpa.br recognition of patterns in their ciphertexts and on the use of machine learning algorithms, with different methods being proposed for identification through the use of classifier algorithms, as in [26][27][28]. However, no identification research involving post-quantum cryptographic algorithms was found. Given this gap, the object of this research is the analysis of cryptograms produced by post-quantum cryptographic algorithms, aiming at the subsequent identification of the generator algorithm, through the use of machine learning algorithms, considering a ciphertext-only scenario, in which only ciphertext samples are found available. Ciphertexts from the post-quantum cryptographic algorithms Frodo, CRYSTALS Kyber, NTRU and Saber – participants in the selective contest implemented by the National Institute of Standards and Technology (NIST) – were analyzed, and useful information extracted from their cryptograms allowed identifying the algorithm’s employees with hit rates that ranged between 73.3% and 100%. Although the scope of this research is post-quantum algorithms, the AES and Blowfish symmetric cryptography algorithms were also analyzed, so that the results obtained could be compared with other results already reported in similar research. II.LITERATURE REVISION There is a wide variety of cryptographic and machine learning algorithms, among which some were used in this research. A. Cryptographic Algorithms Blowfish algorithm was conceived as an alternative to the Data Encryption Standard (DES), due to this algorithm's vulnerabilities to brute force attacks. In [6], Nie, Song and Zhi analyzed the processing speeds and energy consumption of these two algorithms and concluded that Blowfish is significantly faster than DES and that both have similar energy consumption. In research [7], it was concluded that Blowfish provides greater security than Advanced Encryption Standard (AES) and 3DES, due to the key sizes used. Poonia and Yadav analyzed different configurations of the Blowfish algorithm in [8], and presented changes that made it more secure and compact than its original implementation. The Rijndael block cipher was the winner of the selective competition organized by NIST, between January 1997 and October 2000, which instituted the Advanced Encryption Standard (AES) and replaced DES, in accordance with FIPS Post-quantum cryptographic algorithm identification using machine learning B.S. Rocha, J. A. M. Xexéo and R. H. Torres