| 引用本文: | 肖相青， 王元恺， 胡进峰， 刘 军， 钟 凯， 赵紫薇， 李会勇. 一种基于深度学习残差网络的模糊函数赋型方法[J]. 雷达科学与技术, 2024, 22(6): 613-619.[点击复制] |
| XIAO Xiangqing, WANG Yuankai, HU Jinfeng, LIU Jun, ZHONG Kai, ZHAO Ziwei, LI Huiyong. An Ambiguity Function Shaping Method Based on Deep Learning Residual Network[J]. Radar Science and Technology, 2024, 22(6): 613-619.[点击复制] |
|
| 摘要: |
| 基于模糊函数（Ambiguity Function， AF）赋型的恒模波形设计是雷达系统中的一项关键技术。该问题可构造为一个非线性的复四次问题（NP?hard）。现有的方法可分为两类：第一类方法通过松弛方式来求解该问题，但不可避免地会引入近似误差；第二类方法直接求解该问题，但该类方法的参数选取较为困难。我们注意到深度神经网络是一个天然的非线性系统，与上述的非线性问题模型高度契合。因此，本文提出了一种基于深度学习残差网络的方法来对AF赋型，该方法不需要松弛操作以及复杂的参数选取。具体步骤为：1）将该问题转化为一个无约束的相位优化问题；2）将该无约束问题的非凸目标函数构造为网络的损失函数；3）使用残差网络直接优化波形的相位。仿真结果表明，所提方法的信干比（Signal?to?Interference Ratio, SIR）有显著提升并且有着更好的目标探测性能。 |
| 关键词: 深度学习 模糊函数 残差网络 恒模约束 波形设计 |
| DOI：DOI:10.3969/j.issn.1672-2337.2024.06.004 |
| 分类号:TN958.2 |
| 基金项目:国家自然科学基金（No.62231006）； 国家重点研发计划（No.2023YFF0717400）； 衢州市财政资助科研项目（No.2023D040，2023D009，2022D009，2022D013和2022D033）； 四川省科技计划项目（No：2023YFG0176） |
|
| An Ambiguity Function Shaping Method Based on Deep Learning Residual Network |
|
XIAO Xiangqing, WANG Yuankai, HU Jinfeng, LIU Jun, ZHONG Kai, ZHAO Ziwei, LI Huiyong
|
|
1. University of Electronic Science and Technology of China, Chengdu 611731, China;2. The Yangtze Delta Region Institute （Quzhou）, University of Electronic Science and Technology of China, Quzhou 324000, China;3. The 41st Research Institute of China Electronics Technology Group Corporation, Qingdao 266555, China
|
| Abstract: |
| Unimodular waveform design based on ambiguity function （AF） shaping is a crucial technique in radar systems. This problem is formulated as a nonlinear complex quartic optimization problem （NP?hard）. The existing methods can be classified into two categories: the first one solves the problem by relaxing the original problem, but inevitably introducing approximation errors; the second one solves the problem directly, but the selection of parameters in this category is difficult. We notice that deep neural network is a naturally nonlinear system that is highly compatible with that nonlinear problem. Motivated by this, this paper proposes a method based on deep learning residual network for AF shaping without any relaxation or complex parameters selection. The specific steps are as follows: 1） The problem is transformed into an unconstrained phase optimization problem; 2） The non?convex objective function of the unconstrained problem is constructed as the loss function of the network; 3） The residual network is used to directly optimize the phase of the waveform. The simulation results show that the signal?to?interference ratio（SIR） of the proposed method is improved significantly, and it has better target detection performance. |
| Key words: deep learning ambiguity function residual network constant modulus constraint waveform design |
