MIMO雷达稀疏成像的失配问题研究

发布时间：2018-01-22 18:44

本文关键词： MIMO雷达稀疏成像正交匹配追踪算法观测矩阵失配扰动矩阵相位误差载频偏差网格失配　出处：《中国科学技术大学》2014年博士论文　论文类型：学位论文

【摘要】：MIMO(Multiple input Multiple output,MIMO)雷达是指利用多个发射和接收天线同时对目标进行观测的一种新构型的雷达系统。阵列构型设计和波形分集技术使MIMO雷达能够获得远多于实际物理阵元数目的观测通道和空间自由度。通过对观测通道回波的联合处理,相比于传统成像雷达,MIMO雷达在成像的方位向分辨率、实时性和运动补偿等方面有明显的性能优势。进一步的,为克服信号带宽和系统采样频率在实现高分辨率成像时对雷达系统设计和实现的困难和限制,基于压缩感知(Compressed Sensing, CS)的MIMO雷达稀疏成像开始受到广泛的关注,是当前的一个研究热点。由CS理论可知,MIMO雷达的稀疏重构(即,反演)性能依赖于观测矩阵的性质,因此一个精确已知的观测矩阵是获得好的反演结果的前提条件。众所周知,MIMO雷达的观测矩阵由雷达系统参数和成像场景的网格点共同决定,如果其中任一的一个因素存在不确定性都将导致实际观测矩阵不再与默认的观测矩阵一致,这种观测矩阵的失配必然对成像算法的有效性、可靠性和稳健性提出了挑战。因此,研究观测矩阵失配对MIMO雷达稀疏成像的影响是有实际应用意义的。本文采用正交匹配追踪算法(Orthogonal Matching Pursuit,OMP)作为反演算法的比较基准,围绕系统参数和成像场景网格点这两类因素的不确定性,重点研究和分析观测矩阵失配的产生机理、OMP算法在实现有效反演时对这些不确定性的承受能力、以及高效重构算法等问题,主要的研究内容如下： 1、针对相位分集和频率分集两种波形分集方式,建立了对应紧凑式MIMO雷达系统的回波模型,分别从点扩散函数和空间谱的角度推导了成像分辨率和无模糊距离的解析表达式,重点分析了两种角度下对成像分辨率描述的差异。详细介绍了OMP算法的算法流程和基于互相关系数的重构性能推导过程。同时,根据互相关系数和点扩散函数之间的紧密联系,确定了通过点扩散函数来分析观测矩阵失配和稀疏反演性能的可行性。 2、对于系统可能存在的发射-接收通道随机相位误差,基于其在回波相位中不与散射点坐标信息耦合的先验假设,在MIMO雷达系统中建立了含有相位不确定性的回波模型,分析了这一类随机相位误差对观测矩阵的作用形式,表现为一左乘对角扰动矩阵。进一步的,利用受扰动的点扩散函数和相位误差的随机特性,分析了左乘扰动矩阵对OMP算法成像的影响,主要表现为幅度衰减且衰减程度由相位的波动范围决定。特别地,根据推导的OMP算法重构性能,分别在支撑集恢复和幅值估计两方面推导了OMP算法对相位误差的容限。考虑到回波中随机相位误差是一隐含变量的事实,引入期望最大化(Expectation Maximization, EM)方法,根据最大后验概率准则,提出了期望最大化的稀疏成像算法(Sparse Imaging via EM, SIEM),仿真结果显示在存在相位误差时SIEM比OMP具有更稳定的反演性能。 3、对于系统可能存在的发射一接收通道载频偏差,在相位分集MIMO雷达系统中建立了含有发射、接收载频不确定性的解析回波模型,回波表达式表明载频偏差不仅在回波相位中与散射点位置信息强耦合,而且会影响通道分离的性能,导致通道分离残差的出现。相比随机相位误差,载频偏差引起更加复杂、严重的观测矩阵失配。根据受扰动点扩散函数的峰值变化,分析得到了载频偏差对OMP算法成像的影响集中表现为对点扩散函数峰值的衰减,然后进一步推导了存在载频偏差时OMP算法的反演性能变化以及OMP算法支撑集恢复和幅值估计对载频偏差的容限。通过将载频偏差引起的观测矩阵失配表示为一个具有有界Frobenius范数约束的加性扰动矩阵,提出了基于有界扰动的稀疏成像算法(Sparse Imaging based on Frobenius-nrom-bounded Perturbation, SIFrobP)。根据有界扰动的一般性假设,SIFrobP算法的适用范围广泛,可适用于实际观测矩阵中存在任意未知不确定性的场景。 4、研究了连续成像场景的离散化网格与真实目标散射点之间存在不确定性时的网格失配问题。从细化网格提高散射点位置估计精度的角度,将基于Band-exclusion技术的改进型OMP算法(Band-excluded OMP,BOMP)引入MIMO雷达稀疏成像,利用点扩散函数指导相关带门限值的设置使BOMP算法成像的低分辨率得到了有效地改善。同时,从摒弃对连续成像场景网格化的角度出发,提出了基于连续参数估计的MIMO雷达稀疏成像方法(Sparse Imaging via Continuous Parameter Estimate,SICPE),推导了算法的性能条件。该算法不仅避免了经典稀疏重构算法对网格的依赖性,而且可以在发射／接收端稀疏布阵或非均匀采样时均获得较好的稀疏成像结果。
[Abstract]:MIMO (Multiple input Multiple output, MIMO) radar refers to a new type of radar system using multiple transmit and receive antennas at the same time to observe the target. The array configuration design and waveform diversity MIMO radar can obtain much more than the actual number of the array observation channel and spatial degrees of freedom. The combined treatment of the observation channel echo, compared to the traditional imaging radar, MIMO radar resolution in the azimuth, has obvious advantages in real-time and motion compensation. Further, in order to overcome the bandwidth of the signal and system sampling frequency in the realization of high resolution imaging radar system to design and realize the difficulties and limitations, based on compression perception (Compressed Sensing, CS) MIMO radar sparse imaging began to receive widespread concern. It is a research hotspot. According to the CS theory, the sparse MIMO radar Structure (i.e., inversion) performance depends on the nature of the observation matrix, so a precisely known observation matrix is a prerequisite for a good inversion result. As everyone knows, the observation matrix of MIMO radar is determined by the radar system parameters and imaging scene of grid points, if a factor in the existence of any uncertainty will cause the actual observation matrix and observation matrix is no longer consistent by default, this observation matrix mismatch and necessity of imaging algorithm, challenges the reliability and robustness. Therefore, research on the observation matrix is has practical significance of the effect of mismatch between MIMO radar sparse imaging.
This paper uses orthogonal matching pursuit algorithm (Orthogonal Matching Pursuit, OMP) as the benchmark inversion algorithm, surrounding the two kinds of factors of system parameters and the imaging scene grid point uncertainty, focuses on the research and analysis of the observation matrix mismatch mechanism, OMP algorithm to achieve effective inversion on these uncertainty capacity problems and efficient reconstruction algorithm, the main research contents are as follows:
1, considering the phase and frequency diversity of two kinds of waveform diversity, established the corresponding echo model of compact MIMO radar system, which spread function and spatial spectrum angle is deduced analytic expressions of imaging resolution and unambiguous distance from the point, the paper introduced two kinds of description on the imaging resolution angle of the differences in detail. Introduces the OMP algorithm and the reconstruction performance based on derivation of correlation coefficient. At the same time, according to the cross-correlation coefficient and point spread function between close contact, determine the feasibility analysis to the observation matrix mismatch and sparse inversion performance by point spread function.
2, the system may exist for transmitting and receiving channel random phase error, based on the echo phase in scattering and coordinate information coupling hypothesis in MIMO radar system, established a model containing echo phase uncertainty, analyzed this kind of random phase error function to form the observation matrix, performance as a left multiplication diagonal perturbation matrix. Further, based on the characteristics of random disturbance point spread function and phase error, analyzes the impact of the matrix on the left by the disturbance OMP imaging algorithm, mainly for the amplitude attenuation and the attenuation degree is determined by the phase fluctuation. Especially, according to the OMP algorithm performance is. Set recovery and amplitude estimation two derived tolerance OMP algorithm for phase error in support. Considering the random phase error is a latent variable echo in fact, into the expectation maximization (Ex Pectation Maximization (EM) method, according to the maximum a posteriori criterion, proposes the expectation maximization sparse imaging algorithm (Sparse Imaging via EM, SIEM). The simulation results show that SIEM has more stable performance than OMP in the presence of phase error.
3, the system may exist for the launch of a receiving channel in the carrier frequency offset, phase diversity MIMO radar system is established with transmitting, receiving and parsing the echo model of uncertainty of carrier frequency, carrier frequency offset echo expression shows that not only in the echo phase and scattering point position information and strong coupling, and will affect the performance of channel separation, resulting in channel the separation of residuals. Compared with the random phase error, carrier frequency deviation caused by the more complex, the observation matrix serious mismatch. According to the peak point spread function disturbance changes, analysis of the performance effect of carrier frequency offset on OMP algorithm for imaging attenuation of the PSF peak, then the carrier frequency offset are deduced when the change of inversion the performance of the OMP algorithm and OMP algorithm support recovery and amplitude estimation of carrier frequency deviation tolerance. The observation matrix caused by carrier frequency offset The mismatch is expressed as a perturbation matrix with additive bounded Frobenius norm constraint, is proposed based on sparse imaging algorithm of bounded disturbances (Sparse Imaging based on Frobenius-nrom-bounded Perturbation, SIFrobP). According to the general assumption of bounded disturbances, the scope of SIFrobP algorithm is widely applicable to the actual observation matrix in the presence of arbitrary the uncertainty in the scene.
4, there is mismatch between grid uncertainty on continuous imaging scene discretization grid and real target scattering points. From the angle of grid refinement to improve the estimation accuracy of scattering points, based on improved OMP algorithm based on Band-exclusion Technology (Band-excluded OMP BOMP) into MIMO radar imaging using sparse, point spread function guide with threshold value setting to effectively improve the low resolution BOMP imaging algorithm. At the same time, starting from the abandon of continuous imaging scene grid point of view, put forward the MIMO radar imaging method for the sparse parameter estimation based on (Sparse Imaging via Continuous Parameter Estimate, SICPE), derived the performance conditions of the algorithm. This method not only avoids the dependence on the grid of the classic sparse reconstruction algorithm, but also in the transmitting and receiving end or sparse array non uniform sampling are Better sparse imaging results were obtained.

【学位授予单位】：中国科学技术大学
【学位级别】：博士
【学位授予年份】：2014
【分类号】：TN958

【参考文献】