摘要
以Vigenère密码为代表的古典密码算法仍活跃在云计算和在线短信服务等资源受限场景下,以保障信息的安全。近年来,深度学习技术在密码分析领域初现其独特的优势,而其分析机理还需进一步探索。利用深度学习的机器翻译技术,对Vigenère密码变型算法进行了模拟与分析。首先,提出了一种适用于资源受限设备的改进的Vigenère密码变型算法,利用密文频率分布和重合指数验证了其安全性。其次,基于长短时记忆网络和门控循环单元网络实现了密钥长度达35位字符的Vigenère密码变型算法的模拟,并验证了神经网络模型加解密的高效性。进一步建立了密钥恢复模型,实现了Vigenère密码变型算法30位字符的密钥恢复。最后,给出了神经网络超参数的选择建议。
Classical ciphers represented by Vigenere cipher are still active in resource-constrained scenarios such as cloud computing and online SMS services to ensure security.In recent years,deep learning technology has shown its unique advantages in the field of cryptanalysis and its analysis principle needs to be further explored.This paper simulates and analyzes advanced Vigenere cipher with machine translation technology of deep learning.First,this paper proposes an advanced Vigenere cipher algorithm suitable for resource-constrained devices,and its security is verified by using the frequency distribution of ciphertext and the coincidence index.Then,based on long short-term memory(LSTM)networks and gated recurrent unit(GRU)networks,the simulation of ad-vanced Vigenere cipher with a key length of 35 characters is realized,and the efficiency of encryp-tion and decryption of the neural network model is verified.Furthermore,a key recovery model is proposed to realize 30-character key recovery attack of advanced Vigenere cipher.Last,some sug-gestions are presented for the selection of neural network hyperparameters.
作者
林东东
任炯炯
陈少真
LIN Dongdong;REN Jiongjiong;CHEN Shaozhen(Information Engineering University,Zhengzhou 450001,China)
出处
《信息工程大学学报》
2023年第2期215-223,共9页
Journal of Information Engineering University
基金
数学工程与先进计算国家重点实验室开放基金资助项目(2019A08)。
