Block Cipher Identification Scheme Based on Hamming Weight Distribution
Lun Zhao, Yaping Chi, Zihan Xu, Ziyan Yue
Abstract
In order to solve the problem of worse recognition performance under the multi-classification scenarios of existing cryptosystem identification, this paper proposes a block cryptosystem recognition scheme based on Hamming weight distribution. A feature extraction method based on Hamming weight distribution is designed to extract the cipher text features, and the cryptosystem is further described. XGBoost-RFE feature selection algorithm is proposed to filter invalid features. In order to better adapt to the characteristics of complex ciphertext data and difficult fitting in multi classification scenarios, XGB-LGBM ensemble learning model with multi-layer fusion structure is designed to further improve the accuracy and generalization of recognition. The experiment carried out mixed identification of 10 common block cipher algorithm, and the overall recognition accuracy reached 89.65%.