基于RSETransformer的低截获概率雷达信号增强  

LPI radar signal enhancement based on RSETransformer

作　　者：苏琮智 吴宏超[1] 杨承志[1] 邴雨晨 易仁杰 Su Congzhi;Wu Hongchao;Yang Chengzhi;Bing Yuchen;Yi Renjie(Aviation University of Air Force,Changchun 130022,China)

机构地区：[1]空军航空大学,长春130022

出　　处：《战术导弹技术》2022年第5期44-54,140,共12页Tactical Missile Technology

基　　金：国防科技卓越青年科学基金资助项目(315090303)。

摘　　要：针对低截获概率(LPI)雷达发射信号功率低,非合作接收机难以开展后续检测识别任务的问题,提出一种基于Transformer模型的雷达信号增强方法。设计嵌入层对信号进行分块处理,结合多头注意力机制(MSA)建立RSETransformer增强模型,根据信号增强评价指标对损失函数进行设计,通过网络训练隐式地挖掘出带噪雷达信号和纯净雷达信号之间的非线性映射关系,得到包含最优参数的模型实现对信号的增强。通过仿真实验,在低信噪比(SNR)条件下也取得了较优异的增强效果,证明了提出算法的可行性和有效性,能够支撑后续的检测和识别任务。Aiming at the problem that the transmitting signal power of Low Probability of Intercept(LPI)radar is low and it is difficult for non-cooperative receivers to carry out subsequent detection and identification tasks, a radar signal enhancement method based on Transformer model is proposed. Design the Patch Embedding layer to process the signal in blocks and combine the Multi-head Self-Attention(MSA) mechanism to build the RSETransformer enhancement model and design the loss function according to the signal enhancement evaluation index to implicitly find the difference between the noisy radar signal and the pure radar signal through network training. A nonlinear mapping relationship is set up to obtain a model containing the optimal parameters to enhance the signal. Through simulation experiments, excellent enhancement effects are also achieved under the condition of low Signal-to-Noise Ratio(SNR), which can prove the feasibility and effectiveness of the proposed algorithm and support subsequent detection and recognition tasks.

关 键 词：低截获概率雷达 雷达信号增强 Transformer模型 多头注意力机制 RSETransformer 低信噪比 

分 类 号：TN971[电子电信—信号与信息处理]