基于欺骗中继技术的无人机主动监听优化方法研究  

Research on the Optimization Method of UAV Proactive Eavesdropping Based on the Spoofing Relay Technique

作　　者：王贤明 杨超群 曹向辉[2] 龚成龙 张恒 WANG Xianming;YANG Chaoqun;CAO Xianghui;GONG Chenglong;ZHANG Heng(School of Electronic Engineering,Jiangsu Ocean University,Lianyungang 222000,China;School of Automation,Southeast University,Nanjing 211189,China;School of Computer Engineering,Jiangsu Ocean University,Lianyungang 222000,China)

机构地区：[1]江苏海洋大学电子工程学院,连云港222000 [2]东南大学自动化学院,南京211189 [3]江苏海洋大学计算机工程学院,连云港222000

出　　处：《无人系统技术》2024年第3期67-74,共8页Unmanned Systems Technology

基　　金：国家自然科学基金(61873106,62303109)。

摘　　要：针对非法分子通过无线通信危害国家安全的问题,研究了基于无人机欺骗中继技术的合法监听方案,对地面可疑节点之间的通信链路进行监听。首先,将节点之间的链路视为视距链路,对各个信道进行建模,构建了监听率最大化的问题。其次,为了解决这个复杂的非凸优化问题,采用深度强化学习方法,综合考虑无人机的三维轨迹、放大系数和功率分配比这三方面对监听率的影响,将该问题建模为马尔可夫决策过程,设计了相应的奖励函数。最后,基于双延迟深度确定性策略梯度算法实现联合优化。从数值结果来看,相较于基于深度确定性策略梯度算法的主动监听优化策略,所提出的优化策略收敛速度更快,所得到的监听性能有所提升。To address the problem of illegals endangering national security through wireless communications,the paper investigates a lawful eavesdropping scheme based on Unmanned Aerial Vehicle(UAV)spoofing relay technology to eavesdrop on the communication links between suspicious nodes on the ground.Firstly,the problem of maximizing the eavesdropping rate is constructed by considering the link between nodes as a line-of-sight link and modeling each channel.Secondly,to solve this complex non-convex optimization problem,the paper adopts a deep reinforcement learning method,comprehensively considers the impact of the three-dimensional trajectory of the UAV,the amplification coefficient,and the power allocation ratio on the eavesdropping rate,and models the problem as a Markov Decision Process,and designs the corresponding reward function.Finally,the joint optimization is implemented using the Twin Delayed Deep Deterministic Policy Gradient(TD3)algorithm.From the numerical results,compared with the active eavesdropping optimization strategy based on Deep Deterministic Policy Gradient algorithm,the optimization strategy based on the TD3 algorithm proposed in this paper has a faster convergence speed,and the performance of eavesdropping is improved.

关 键 词：无人机 深度强化学习 欺骗中继 合法监听 监听速率 

分 类 号：TN929.5[电子电信—通信与信息系统]