Research On GAN Anti-encryption Method Based On Chaos Model

Today is an era with information developing rapidly.The Internet and IT technologies are continuously popularized and used in various industries.Information is very easy to be stolen or spread in the network transmission,which leads to many security problems.Therefore,the research on digital information security is inevitable,and in the long run,its application prospect is broad and its application value is increasing day by day.Then how to protect information security? One of the effective ways is password,which encrypts the data when it is transmitted to ensure the security of information.Neural networks have a characteristic that is particularly like encryption in traditional cryptography,that is highly nonlinear dynamic system.As a result of this,neural networks have a broad research prospect,which inspires many scholars to combine them with chaos to predict random sequences and apply them in the field of encryption.In this paper,the encryption method combining neural networks and traditional cryptography is studied.Introducing the adversarial attack into traditional cryptography by analyzing the idea of generative adversarial network and the security of adversarial encrypted communication model,and proposing a kind of adversarial encryption method which use generative adversarial network and based on chaos model.Firstly,the security of the adversarial encrypted communication model is tested by endowing the attacker with stronger attack ability,and the structure of the model is improved.Secondly,using GAN to generate chaotic random sequence as encryption keys,and input the keys generated by GAN into the improved adversarial encrypted communication model.A secure encryption method is obtained through the adversarial training among encryptor,decryptor and attacker.The encryption method has a good effect on eliminating the periodicity of chaos,Moreover,it can enhance the ability of ciphertext to resist attacks and ensure the security of data effectively.