宽带噪声雷达压缩感知成像方法研究

Noise radar has the advantage of excellent resistance to jamming and interference, low probability of intercept and low probability of detection, but the application of noise radar in (inverse) synthetic aperture radar imaging has the disadvantage of high-level sidelobes and low dynamic range. The application of compressed sensing (CS) theory in radar imaging could reduce the data, but it is confronted with the difficulty of designing complicated sampling devices, thus the imaging system and method for direct undersampling at radar receiver is needed. Our recent investigations find that, these drawbacks could be overcome by combining the noise radar technology and the CS theory. This project contains two aspects. First, starting from investigating the connections between noise radar and the random measuring strategy in CS, we will construct the sparse optimization model for suppressing the masking effect in noise radar based on sparse representations and sparse recovery methods, and thus improve the image quality of conventional noise radar. Second, this project will study the CS-based imaging methods for noise radar using low-rate A/D sampled data and low-bit quantized data, respectively. New mathematical model and optimization algorithm will be proposed to reduce the complexity of sampling devices and the size of data for wideband radar. Imaging experiments of noise radar will be carried out based on present radar systems, and the obtained data will be used to evaluate and improve theoretical models and algorithms. Finally, a complete theoretical CS-based noise radar imaging framework will be utilized to promote the practical application of noise radar and CS radar imaging.

噪声雷达具有抗干扰、低检测和低截获的优点，但在(逆)合成孔径雷达成像时存在旁瓣水平高、动态范围低的缺点。压缩感知理论应用于雷达成像可以降低数据量，但面临设计复杂采样设备的问题，欠缺直接在接收端降采样的雷达成像系统和方法。我们前期研究发现，将噪声雷达技术和压缩感知理论结合起来，有望弥补二者的不足。本项目包括两个内容：(1) 从研究噪声信号与压缩感知随机测量的关系入手，基于压缩感知建立抑制噪声雷达mask效应的稀疏优化模型，提升传统噪声雷达的成像质量；(2) 针对低速A/D采样和低比特量化两种数据获取方式，分别研究噪声雷达的压缩感知成像方法，通过设计新的数学模型和优化算法，降低宽带雷达成像的采样难度和数据量。本项目将基于现有雷达实验系统开展噪声雷达成像实验，通过对实际观测数据的处理评估和改进模型和算法，最终建立一个完整的基于压缩感知的噪声雷达成像理论框架，推动噪声雷达和压缩感知成像的实际应用。

噪声雷达具有的抗干扰、低检测和低截获的优势，但在成像领域面临着旁瓣高、动态范围低的问题。将压缩感知与噪声雷达结合起来，可以提升噪声雷达的成像质量，同时降低采样率和数据量。针对宽带噪声雷达的压缩感知成像方法，课题开展了以下研究：.(1) 研究了噪声雷达波形设计问题，分析了噪声波形与压缩感知测量矩阵相关性的联系，提出了一种低相关性、低旁瓣的噪声雷达信号模型。通过在频域调制随机雷达信号的频谱来实现低旁瓣的效果，避免了随机雷达信号过高的旁瓣造成的掩盖效应。.(2) 研究了低速ADC采样下的噪声雷达波形设计方法，提出了一种新型线性调频随机雷达信号模型及其去斜处理方法，兼具了随机雷达信号和线性调频信号的优点，还可采用去斜处理方法实现低采样率下脉冲压缩，降低了对雷达系统ADC硬件采样率的要求。.(3) 研究了低速ADC采样下的宽带噪声ISAR成像方法，针对频率步进噪声信号，设计了压缩感知距离压缩方法，能够从低于奈奎斯特采样率下的ADC采集数据中恢复信号。在此基础上开展了运动目标的ISAR成像实验，以1/2奈奎斯特采样率的ADC采集数据，成功得到高分辨率的ISAR图像。.(4) 研究了噪声雷达图像增强算法，针对噪声雷达高旁瓣导致的掩盖效应，提出了稀疏优化算法和非匹配滤波算法。对实际机载噪声SAR的数据处理表明，两种方法都可以有效抑制噪声雷达的掩盖效应，增强图像动态范围。.(5) 研究了低比特量化下的噪声雷达成像算法，在不同信噪比下对比分析了线性调频信号和噪声信号的单比特脉冲压缩结果，线性调频信号单比特量化会出现虚假目标，噪声信号单比特量化只会导致背景噪声增加，不会出现虚假目标。