| With the improvement of computer computing power and the development of cryptography technology,password cracking technology is also constantly upgrading.Classical cryptography mainly studies traditional cryptographic algorithms,such as Caesar ciphers,single table substitution ciphers,etc.These algorithms have relatively low security and can be cracked through simple plaintext cracking methods.Modern cryptography mainly studies more complex encryption algorithms,such as Advanced Encryption Standard(AES)algorithm,Data Encryption Standard(Data Encryption Standard)Data Encryption Standard,etc.These algorithms have high security and need more complex cracking methods.Nowadays,technologies such as artificial intelligence and machine learning can be used to improve the efficiency and accuracy of plaintext cracking.A neural network is a computational model that simulates the human brain,with adaptability and learning ability.In the field of cryptography,neural network technology can be used for password generation,encryption,decryption,attacks,and other aspects.This article uses neural network technology to conduct plaintext guessing on symmetric encryption algorithms,attempting to use the network model constructed by neural networks to guess unknown plaintext data in the encryption algorithm,and proposing new ideas for password cracking.The research of this article mainly includes two contents:(1)Propose a symmetric plaintext guessing method based on a combination convolutional neural network model.It is mainly aimed at four popular symmetric encryption algorithms,namely:Advanced Encryption Standard AES algorithm,Data Encryption Standard Data Encryption Standard,symmetric block encryption Blow fish algorithm and stream cipher RC4 algorithm.The specific work is to design a composite convolutional neural network model constructed by the Tensorflow framework,where the convolutional neural network is composed of a fully connected layer and four one-dimensional convolutional layers.Then,the model was used to guess unknown plaintext in specific scenarios,and the experimental results showed that the guess hit rate was at least 60%.(2)Propose a plaintext guessing method for SM4 ciphers based on a combined convolutional neural network model.Mainly aimed at the SM4 algorithm in the national security algorithm,which is a domestically produced symmetric encryption algorithm with strong password security.The combination convolutional neural network model proposed in the previous chapter was used to guess its formation.The experimental results showed that it was not ideal.Compared with the previous symmetric encryption algorithm,the SM4 algorithm introduced more nonlinear transformations.Considering that the neural network is not good at summarizing the characteristics of non-linear factors,this may be one of the reasons for its poor guess performance,Subsequently,the impact of nonlinear transformations in neural networks on the model was discussed,and the shortcomings of the model were analyzed.(3)Based on the methods proposed in the first two items,this article also designs a guessing model system for symmetric encryption algorithms.The system includes four modules,namely the selection algorithm module,algorithm encryption module,model guessing module,and comparison data module.The four modules use step-by-step operations to transfer data,and finally complete the plaintext guessing work of the symmetric encryption algorithm,This provides a guarantee for the visualization of the guessing model of future asymmetric encryption algorithms,and can promote a more comprehensive understanding of the significance of guessing work. |
