A Block Cipher Algorithm Identification Scheme Based on Hybrid Random Forest and Logistic Regression Model

Cryptographic algorithm identification is aimed to analyze the potential feature information in ciphertext data when the ciphertext is known, which belongs to the category of cryptanalysis. This paper takes block cipher algorithm as the research object, and proposes a block cipher algorithm identification scheme based on hybrid random forest and logistic regression (HRFLR) model with the idea of ensemble learning. Based on the NIST randomness test feature extraction method, five block ciphers, AES, 3DES, Blowfish, CAST and RC2, are selected as the research object of cryptographic algorithm identification to carry out the ciphertext classification tasks. The experimental results show that, compared with the existing methods, the cryptographic algorithm identification scheme based on HRFLR proposed in this paper has higher accuracy and stability on binary classification and multi-class classification tasks. In the binary classification tasks of AES and 3DES, the identification accuracy of our proposed cryptographic algorithm identification scheme based on HRFLR can reach up to 74%, and the highest identification accuracy of the five classification tasks is 38%. Compared with the 54% and 28.8% accuracies of random forest-based identification scheme, the accuracy is increased by 37.04% and 18.06%, respectively. This result is significantly better than the 50% and 20% accuracies of random guessing scheme.