基于深度神经网络的网络信息数据传输自动加密算法研究  

Research on automatic encryption algorithm of network information data transmission based on deep neural network

作　　者：宗满贵 任晓龙 ZONG Mangui;REN Xiaolong(CETC(Beijing)Network Information Security Co.,Ltd.,Beijing 100070,China;Institute of Computing Technology,Chinese Academy of Sciences,Beijing 100190,China;University of Chinese Academy of Sciences,Beijing 100190,China)

机构地区：[1]中电科(北京)网络信息安全有限公司,北京100070 [2]中国科学院计算技术研究所,北京100190 [3]中国科学院大学,北京100190

出　　处：《电子设计工程》2024年第19期90-93,共4页Electronic Design Engineering

摘　　要：为更好地应对各种复杂和多变的安全威胁,提出基于深度神经网络的网络信息数据传输自动加密算法。利用级联Chebyshev映射,生成网络信息数据的伪随机混沌序列;通过生成对抗网络处理伪随机混沌序列,生成网络信息数据的置乱矩阵;通过长短期记忆网络处理混沌序列,生成网络信息数据的置乱矩阵;利用加权融合方式,融合两个网络生成的置乱矩阵;按照融合后的置乱矩阵置乱处理网络信息数据,完成网络信息数据传输自动加密。实验证明,在承受复杂网络攻击时,该算法具备较优的自愈性能,提升了网络信息数据传输的安全性。In order to better deal with various complex and changeable security threats,an automatic encryption algorithm of network information data transmission based on deep neural network is proposed.The pseudo random chaotic sequence of network information data is generated by cascade Chebyshev mapping.By generating adversarial networks to process pseudo-random chaotic sequences,the chaotic matrix of network information data is generated.The chaotic sequence is processed by long and short term memory network,and the scrambling matrix of network information data is generated.The weighted fusion method is used to fuse the scrambling matrix generated by the two networks.The network information data is scrambled according to the scrambled matrix after fusion,and the network information data transmission is automatically encrypted.Experiments show that in the case of complex network attacks,the algorithm has better self-healing performance,and improves the security of network information data transmission.

关 键 词：深度神经网络 网络信息 数据传输 自动加密 CHEBYSHEV映射 置乱矩阵 

分 类 号：TN914[电子电信—通信与信息系统]