【作者】 赵霞；

【导师】 郭陈江；

【作者基本信息】 西北工业大学 ， 电子科学与技术， 2019， 博士

【摘要】 双基地MIMO（Multiple-input and Multiple-output）雷达利用一组发射阵元发射相互正交的信号,一组接收阵元采集目标回波信号。由于利用空间分集可以形成大的虚拟阵列孔径,能有效地对目标进行探测和定位。为了实现对双基地MIMO雷达目标的定位,论文从空间谱估计、相关信号处理、发射信号设计和稀疏信号重构角度进行了研究。主要研究成果可概括为:基于子空间原理提高了线性预测算法对双基地MIMO雷达目标定位的精度,基于相关矩阵重构实现了相关信号下双基地MIMO雷达目标定位,基于J正交矩阵构造发射信号提高目标定位精度,从稀疏信号重构角度实现双基地MIMO雷达目标三维参数联合估计以及采用改进l₂范数优化算法实现密集网格下的目标快速定位。1.研究了双基地MIMO雷达的信号模型及相关理论。基于闵可夫斯基加法和集合元素减法讨论了虚拟阵列的构成。在发射正交波形和匹配滤波的前提下,从电磁波发射、反射和接收过程,详细推导了双基地MIMO雷达单目标信号模型,并推广到多目标信号模型。根据克拉美劳界的定义,推导了双基地MIMO雷达单目标三参数和多目标三参数估计时的克拉美劳界。利用子空间原理,对线性预测算法进行改进,解决了多参数估计时,线性预测算法应用到双基地MIMO雷达目标定位时精度不高的问题。2.采用反对角阵加载和酉变换,实现了双基地MIMO雷达目标的DOD（Direction of Departure）和DOA（Direction of Arrival）自动匹配估计,同时提高了目标定位精度,降低了计算复杂度。通过自相关矩阵和互相关矩阵求解多普勒频率,当多普勒频率相同时会出现相关问题,造成自相关矩阵秩亏损,无法直接通过本征值分解对双基地MIMO雷达目标进行定位。因此基于矩阵重构思想,采用对角前向空间平滑法DFSS＿ESPRIT（Diagonal Forward Space Smoothing＿Estimating Signal Parameters via Rotational Invariance Techniques）、截断对角前向空间平滑法TDFSS＿ESPRIT（Truncate DFSS＿ESPRIT）、基于Toeplitz矩阵的Toep＿ESPRIT法和差分空间平滑法DSS＿ESPRIT（Differential Space Smoothing＿ESPRIT）,解决了自相关矩阵秩亏损问题,实现了信号相关时的双基地MIMO雷达目标定位。3.采用J正交矩阵构造发射信号,提高了双基地MIMO雷达目标的定位精度。结合J正交矩阵的定义,基于主枢轴变换,分析了J正交矩阵具有的性质。仿真分析了单位阵、高斯阵、Hadamard阵和J正交矩阵构造发射信号时,对双基地MIMO雷达目标角度克拉美劳界的影响。4.基于稀疏信号重构算法,实现了双基地MIMO雷达目标三维参数的联合估计。传统的谱估计算法基于空间谱,所以只能同时估计出目标的二维参数。与之不同,稀疏重构算法通过双基地MIMO雷达稀疏信号模型,将空间谱中的谱峰搜索转变为完备字典中反射系数的峰值搜索,实现了三维参数估计。采用奇异值分解解决了奇异矩阵求逆问题,并用残差信号控制迭代步长的自适应Tikihonov算法同时估计出目标的波达方向、波离方向和反射系数,并且每个目标的各参数自动匹配。5.采用加权l₂范数算法,实现了密集网格下双基地MIMO雷达目标三维参数的联合估计,解决了密集网格下的定位难题。密集网格划分虽然有助于提高目标定位范围,但是数据量大,算法收敛速度缓慢。为了解决这一问题,改变约束条件构建新的目标函数,并采用嵌套共轭梯度的方法,从稀疏信号重构角度实现了密集网格下,双基地MIMO雷达目标的波达方向、波离方向和反射系数快速估计,并且目标各参数自动匹配。更多还原

【Abstract】 Bistatic multiple-input and multiple-output（MIMO）radar utilizes a group of transmitting antennas to transmit orthogonal signals,and a group of receiving antennas to collect echo signals from targets.With a large virtual array aperture formed by spatial diversity,bistatic MIMO radar can effectively detect and locate targets.In order to achieve targets location of bistatic MIMO radar,the spatial spectrum estimation algorithm,coherent signal processing,transmission signal design and sparse signal restoration are studied.The main research results can be summarized as follows:Linear prediction（LP）algorithm based on the subspace principle to improve the precise of target location,algorithms based on correlation matrix reconstruction to locate targets under coherent surroundings,transmitted signals based on J orthogonal matrix to improve positioning precise,the joint estimation of 3D parameters of targets in bistatic MIMO radar from sparse signal reconstruction and the improved l₂ norm optimization algorithm to achieve fast target location under dense bins.1.The signal model and related theory of bistatic MIMO radar are studied.Virtual array is discussed based on Minkowski addition and set element subtraction.Under the premise of orthogonal transmitting waveform and matched filtering,the single-target signal model of bistatic MIMO radar is deduced from the transmitting,reflecting and receiving process of electromagnetic wave,and is extended to multi-target signal model.According to the definition of Cramer-Rao bound（CRB）,the CRB for three-parameter of single-object in bistatic MIMO radar and the CRB for three-parameter of multi-objective in bistatic MIMO radar are deduced.As far as multi-parameter estimation in bistatic MIMO radar is concerned,the location precision of LP algorithm decreases.The problem is resolved by subspace principle.2.Anti-diagonal matrix and unitary transformation are used to realize the autopaired direction of departure（DOD）and direction of arrival（DOA）estimation of targets in bistatic MIMO radar,with improved target positioning precise and decreased computational complexity.Based on the autocorrelation matrix and the cross-correlation matrix of received signals,multiple Doppler frequencys are estimated.But it is found that the coherent problem emerges when multiple Doppler frequencys are the same,which causes the rank loss of autocorrelation matrix.Under such circumstances,it is impossible to locate bistatic MIMO radar target directly through eigenvalue decomposition（EVD）.Therefore,based on the idea of matrix reconstruction,diagonal forward space smoothing Estimating Signal Parameters via Rotational Invariance Techniques（DFSS＿ESPRIT）,truncate DFSS＿ESPRIT（TDFSS＿ESPRIT）,Toeplitz-based ESPRIT（Toep＿ESPRIT）and differential space smoothing ESPRIT（DSS＿ESPRIT）are used to solve the rank loss of autocorrelation matrix,and the location of target in bistatic MIMO radar is realized.3.J orthogonal matrix is used to construct the transmitted signal,which improves the positioning accuracy of the target in bistatic MIMO radar.Based on the definition of J orthogonal matrix and the principal pivot transformation,the properties of J orthogonal matrix is analyzed.The effects of unit array,Gaussian matrix,Hadamard matrix and J orthogonal matrix on the CRB of target angles in bistatic MIMO radar are simulated and analyzed.4.Based on a sparse signal reconstruction algorithm,the simultaneous estimation of the 3D parameters of target in bistatic MIMO radar is realized.The traditional spatial spectrum estimation algorithm can only realize estimation of 2D parameters at the same time.The sparse signal restoration algorithm estimates 3D parameters by using the sparse signal model of bistatic MIMO radar and by converting peak search in spatial spectrum into the peak search of reflection coefficient in a complete dictionary.The singular value decomposition is used to solve the inversion problem of a singular matrix.Adaptive Tikihonov algorithm with the iterative step controlled by the residual signal is utilized to estimate the direction of arrival,the direction of departure and the reflection coefficient of targets simultaneously,and the multiple parameters of targets are automatically matched.5.A weighted l₂ norm algorithm is used to realize the automatic 3D parameters estimation of targets in bistatic MIMO radar under dense bins.The algorithm solves the positioning problem under dense bins.Although dense bins help to improve the target location range,the amount of data involved is large and the algorithm converges slowly.In order to solve this problem,a constraint condition is used to construct a new objective function,and the conjugate gradient algorithm is used.The direction of arrival,the direction of departure and the reflection coefficient of targets in bistatic MIMO radar under dense bins are estimated from the perspective of sparse signal reconstruction.Parameters of every target are automatically matched.更多还原