神经网络在无线隐蔽通信中的应用
Covert communication based on neural networks
于季弘 1林子砚 1杨传敏 1蔡雨庭 1刘家豪 1王帅1
作者信息
- 1. 北京理工大学 信息与电子学院,北京 100081
- 折叠
摘要
无线通信技术已经应用到社会人、机、物等多种元素中,承载着包含多种隐私数据的无线信号.无线传输信道的开放性使其安全性受到了不断的挑战,无线隐蔽通信技术在实现了通信意图安全的同时保证了信息安全和通信路径安全.该文介绍了无线隐蔽通信系统的经典模型,并总结、归纳了传统方法下的隐蔽性能分析和隐蔽系统设计;介绍了利用对抗神经网络解决不同隐蔽通信场景下的干扰设计、中继功率分配和可重构智能表面设计等问题;以利用公开信号作为掩体的隐蔽通信场景为例,介绍了一种利用生成对抗网络生成有限长隐蔽信号的方案,并进一步给出应用该网络设计全双工接收机的干扰信号;探讨了神经网络方法在中继隐蔽通信、非正交多址接入(NOMA)下的隐蔽通信,瑞丽衰落下的隐蔽通信,干扰辅助的隐蔽通信场景下的应用前景以及进一步的研究方向.
Abstract
Wireless communication technology has been applied to various elements of society,including human-computer interaction and physical objects,carrying wireless signals containing various types of private data.The openness of wireless transmission channels has continuously challenged their security,and wireless covert communication technology ensures in-formation security and communication path security while achieving secure communication intentions.This paper introduces the classic model of covert communication systems and summarizes and generalizes the relevant research of covert communi-cation under traditional methods.This paper also introduces the utilization of adversarial neural networks to design effective jamming,power allocation in relay communication,and reconfigurable intelligent surfaces.Taking the scenario of covert communication with legitimate users as an example,we show a method based on a generative adversarial network to design limited-length covert signals and further apply the network to design interference signals under a full-duplex receiver.Final-ly,we discuss the academic value and potential research directions of neural network methods in relay covert communica-tion,covert communication under non-orthogonal multiple access(NOMA),covert communication with Rician fading,and interference-assisted covert communication.
关键词
无线隐蔽通信/神经网络/机器学习/生成对抗网络Key words
wireless covert communication/neural networks/machine learning/generative adversarial neural networks引用本文复制引用
基金项目
国家自然科学基金项目(62271055)
国家自然科学基金项目(U1836201)
出版年
2024
国家科技期刊平台
NSTL
万方数据