基于深度学习的信息高保密率传输方法  

High Security Information Transmission Technology Based on Deep Learning

作　　者：颜祺 牛彦杰 陈国友 Yan Qi;Niu Yanjie;Chen Guoyou(College of Command and Control Engineering,Army Engineering University,Nanjing 210001;Unit 94860 of the PLA,Nanjing 210046)

机构地区：[1]陆军工程大学指挥控制工程学院,南京210001 [2]中国人民解放军94860部队,南京210046

出　　处：《信息安全研究》2022年第8期793-800,共8页Journal of Information Security Research

基　　金：江苏省重点研发计划项目(BE2018754,BE2019762,BE2020729)。

摘　　要：在信息化、智能化融合发展的时代背景下,多跳无线通信网络作为连接多用户通信设备的关键载体,在进行远距离无线中继传输过程中往往面临着信息被窃听的威胁.为了提升在无线通信过程中的信息安全,提出了一种基于深度学习的信息高保密率传输方法,其核心是通过最优的中继选择来保证信息传输过程的安全性.首先建立人工神经网络模型,利用在多种通信环境下的信道状态数据来训练模型,从而分类选择多跳中继的方案,并获得解码转发(decode-and-forward,DF)中继约束下最大化的保密率.仿真结果表明,相比于传统的穷举搜索方法,该方法可以实现接近0.2 b/(Hz·s)的保密性能.而且,由于模型训练是前期工作,在实际应用中将直接根据信道信息返回结果,因此可显著减少计算时间.Under the background of the integration of information and intelligence,multi-hop wireless communication network,as a key carrier connecting multi-user communication equipment,is often faced with the threat of information eavesdropping in the process of long-distance wireless relay transmission.In order to improve the information security in the wireless communication process,this paper proposes a high security information transmission method based on deep learning,whose core is to ensure the security of the information transmission process through optimal relay selection.Firstly,the artificial neural network model is established,and the channel state data in various communication environments are used to train the model,so as to select the multi-hop relay scheme and obtain the maximum security rate under the decode and forward(DF)relay constraint.Simulation results show that compared with the traditional exhaustive search method,the proposed method can achieve a security performance close to 0.2 b/(Hz·s).Moreover,since the model training is preliminary work,the results will be returned directly according to the channel information in practical application,so the calculation time can be significantly reduced.

关 键 词：无线网络 保密率 中继选择 解码转发 深度学习 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程]