Classification of Encryption Algorithms using Fisher’s Discriminant Analysis

Abstract

<p>Fisher’s Discriminant Analysis (FDA) is a method used in statistics and machine learning which can often lead to good classification between several populations by maximizing the separation between the populations. We will present some applications of FDA that discriminate between cipher texts in terms of a finite set of encryption algorithms. Specifically, we use ten algorithms, five each of stream and block cipher types. Our results display good classification with some of the features. In the present case we have little in terms of an existing standard; however, our limited study clearly shows that further exploration of this issue could be worthwhile.</p>

Cryptanalysis of an Image Cipher using Multi entropy Measures and the Countermeasures

“…Traffic analysis caters the analysis and monitoring of data communicated and communication activities over communication networks. It is an important Task and lot of relevant work is reported in traffic analysis [35][36][37][38][39] which helps in future planning of networks 36 , formulation of practices 37 , network selection 38 , traffic prediction 39 , identifying and classifying communications such as watermarked and noisy images 40 , multimedia and text classification [41][42][43] , engineering of communication networks 44 , and classification of encryption algorithms [45][46] . Identification of cipher data of same key is one of the task of traffic analysis to identify and segregate for extraction of vital intelligible information of adversaries.…”

One to One Identification of Cryptosystem Using Fisher’s Discriminant Analysis

“…It has been used in many areas such as image processing and pattern recognition 12 , machine intelligence 11 , in the classification of speech or music 10 . But few researchers applied FDA into cryptanalysis, only Ray et al in 2017 13 applied FDA in ciphertext classification. 5 kinds of stream ciphers and 5 kinds of block ciphers were classified by 3 kinds of extracted features.…”

“…In order to make up for this deficiency, this paper aims at the improvement of the FDA-based classification technique proposed in 13 , and selects practical 18 algorithms which are 4 kinds of stream ciphers and 7 kinds of block ciphers in ECB mode and CBC mode. After analysis, 9 features completely different from the 3 features in Ref.…”

“…After analysis, 9 features completely different from the 3 features in Ref. 13 were extracted. According to the results of the features extraction, FDA technique was used to identify the cryptosystems one to one.…”

Classifying Classic Ciphers using Machine Learning

“…The paper [13] compares the results from HMMs [12] to Fisher's discriminant analysis (FDA). The papers [12] and [13] collect the data in form of ciphertext's bit information, and so, ciphertext themselves are used as the training data. One similar research [14], uses neural networks to classify three classic substitution ciphertexts along with certain features.…”