Research on Plaintext Resilient Encrypted Communication Based on Generative Adversarial Network  

作　　者：Xiaowei Duan Yiliang Han Chao Wang Xinliang Bi 

机构地区：[1]College of Cryptographic Engineering,Engineering University of PAP,Xi’an 710086,China [2]Key Laboratory of PAP for Cryptology and Information Security,Xi’an 710086,China

出　　处：《国际计算机前沿大会会议论文集》2021年第2期151-165,共15页International Conference of Pioneering Computer Scientists, Engineers and Educators（ICPCSEE）

基　　金：The National Natural Science Foundation of China(No.61572521);Engineering University of PAP Innovation Team Science Foundation(No.KYTD201805);Natural Science Basic Research Plan in Shaanxi Province of China(2021JM252).

摘　　要：Encrypted communication using artificial intelligence is a new challenging research direction.GoogleBrain first proposed the generation of encrypted communication based on Generative Adversarial Networks,resulting in related discussions and research.In the encrypted communication model,when part of the plaintext is leaked to the attacker,it will cause slow decryption or even being unable to decrypt the communication party,and the high success rate of the attacker decryption,making the encrypted communication no longer secure.In the case of 16-bit plaintext symmetrical encrypted communication,the neural network model used in the original encrypted communicationwas optimized.The optimized communication party can complete the decryption within 1000 steps,and the error rate of the attacker is increased to more than 0.9 without affecting the decryption of the communicator,which reduces the loss rate of the entire communication to less than 0.05.The optimized neural network can ensure secure encrypted communication of information.

关 键 词：Generative adversarial networks Artificial intelligence Convolutional neural network Plaintext leakage 

分 类 号：TP3[自动化与计算机技术—计算机科学与技术]