【作者】 王鹏；

【导师】 邱天爽；

【作者基本信息】 大连理工大学 ， 信号与信息处理， 2016， 博士

【摘要】 测向定位是无线被动定位中应用最为广泛的定位技术,其基本任务是实现目标波达方向(DOA)的估计。现有研究多基于高斯噪声和远场源的假设,而在实际应用中,脉冲噪声和阵列的近场效应是普遍存在的。另一方面,基于飞机散射信号的地面干扰源定位新技术很大程度上克服了现有干扰源监测技术的不足。但现有研究多集中在定位算法上,对飞机散射信号的多普勒频移估计研究较少,而多普勒频移的高精度估计是实现地面干扰源定位的前提。因此本文重点针对脉冲噪声下远场源DOA估计、近场源多参数联合估计以及飞机散射信号的多普勒频移估计进行了系统且深入的研究。本文对无线定位中波达方向与多普勒频移估计问题进行了研究,主要贡献如下：(1)针对脉冲噪声下由相关熵矩阵无法得到正交信号子空间和噪声子空间的问题,本文构造了一种基于相关熵的代价函数,利用最大相关熵准则进行约束,设计了一种迭代算法求解优化问题并实现了信号子空间的估计。进一步地,为降低计算复杂度,本文提出了一种新的广义相关熵,证明了对称Alpha稳定分布随机变量广义相关熵的有界性,提出了一种基于最小广义相关熵准则的DOA估计算法。最后,针对短快拍与相干源的DOA估计问题,本文提出了一种最大相关熵准则下基于稀疏表示的DOA估计新算法。仿真结果表明,与现有算法相比,本文所提上述算法在脉冲噪声尤其是高脉冲性噪声下具有非常明显的优势。(2)针对现有基于四阶累积量的近场信源多参数联合估计算法存在计算量大和阵列孔径损失的问题,本文通过构造新的四阶累积量矩阵提出了一种近场源多参数联合估计新算法,避免了阵列孔径损失、降低了计算复杂度并提高了参数估计精度。进一步地,为抑制脉冲噪声,本文提出了一种基于相位分数低阶矩(PFLOM)的近场源多参数联合估计算法,避免了阵列孔径损失、谱峰搜索以及参数配对。为实现参数估计性能的评价,本文首次从理论上分析了柯西噪声下的近场源参数估计的克拉美-罗界。(3)针对非合作双基地雷达系统中时延与多普勒频移联合估计问题,本文在分析线性调频信号(LFM)短时分数阶傅里叶变换(STFRFT)时频特征基础上提出了一种基于STFRFT的联合估计算法,提高了算法的抗噪性。针对飞机散射信号多普勒频移估计问题,本文在分析散射信号多普勒频移特征基础上提出利用LFM信号对短时间内的多普勒频移进行近似,将其转化为LFM信号的参数估计问题。针对低占空比问题,本文提出了高斯加权分数阶傅里叶变换(GFRFT)的概念,对GFRFT的分数阶Fourier域输出信噪比进行了分析,并提出了一种基于GFRFT的多普勒频移估计方法。进一步地,本文分析了对称Alpha稳定分布噪声的分数阶Fourier域特征,提出了PFLOM-FRFT的概念,实现了脉冲噪声下的多普勒频移估计。仿真与实际地面实验结果表明,两种算法均能实现飞机散射信号多普勒频移的高精度估计。更多还原

【Abstract】 Direction finding and location is the most widely used technology in wireless passive location, and its basic task is to realize the estimation of direction of arrival (DOA). Most of the existing researches rely heavily on the assumptions that the noise is Gaussian and the source is located relatively far from the antenna array. However, in some scenarios, the noise exhibits an impulsive nature mainly characterized by sudden bursts or sharp spikes, and the source is located in the near-field of the array aperture. On the other hand, a novel interference source location technology based on the Doppler shift of scattering signal has been put forward, and can overcome the drawbacks of the existing monitoring techniques. However, the existing researches mostly focus on the location algorithm, and very little has been done for the estimation of Doppler shift of scattering signal, which is the premise to realize the interference source location. Therefore, the DOA estimation of far-field source in impulsive noise, the near-field source localization and the Doppler shift estimation of scattering signal will be intensively studied in this dissertation.The DOA and Doppler shift estimation are studied in this dissertation, and the main contributions are listed as follows.(1) In order to overcome the drawback that the orthogonal signal subspace and noise subspace cannot be obtained by the eigen-decomposition of the correntropy-based covariance matrix for impulsive noise, a novel correntropy-based cost function is proposed, and the signal subspace is estimated by solving the optimization problem under the maximum correntropy criterion. To solve the optimization problem, an optimal step size based iterative algorithm is developed and the convergence of this algorithm is proved. Furthermore, to reduce the computational complexity, a generalized correntropy is proposed. The boundedness of the generalized correntropy for symmetric alpha stable (SaS) variable is proved, and a DOA estimator based on the minimum generalized correntropy criterion is proposed. Finally, for the DOA estimation for short snapshots and coherent sources, a novel DOA estimator is proposed by fusing the correntropy and the sparse representation. The simulation results demonstrate that the proposed methods are superior to some existing methods in impulsive noise, especially in highly impulsive noise.(2) To reduce the computational complexity and avoid the loss of array aperture existing in the existing fourth-order cumulant based methods, a novel multi-parameters joint estimation method for near-field source is proposed by constructing new cumulant-based matrices, which avoids the loss the array aperture, reduces the computational complexity and improves the accuracy of parameters estimation. Furthermore, to realize the near-field source localization in impulsive noise, a novel method is proposed by adopting the phased fractional lower-order moment (PFLOM). The algorithm avoids the loss of array aperture, the peak search and the parameter pairing. To characterize its performance, the Cramer-Rao bounds for the estimated parameters of near-field sources in Cauchy noise are derived for the first time.(3) For the joint estimation of time delay and Doppler shift in non-cooperative bistatic radar, based on the feature analysis of the short-time fractional Fourier transform (STFRFT) of linear frequency modulated (LFM) signal, a novel STFRFT-based method is developed in this dissertation. The simulation results demonstrate that the proposed method improves the parameter estimation performance in strong noise environments. For the estimation of Doppler shift of aircraft scattering signal, the Doppler shift is approximated by LFM signal based on the feature analysis of Doppler shift, and the estimation of Doppler shift is converted to the parameter estimation of LFM signal. To improve the performance in low duty ratio condition, a novel Gaussian-weighted fractional Fourier transform (GFRFT) is defined, and the output signal-to-noise ratio in fraction Fourier domain is analyzed. Furthermore, to realize the Doppler shift estimation in impulsive noise, a novel phased fractional lower-order moment based fractional Fourier transform (PFLOM-FRFT) is defined, and the feature of complex SaS noise in fractional Fourier domain is analyzed. Both simulation and actual ground experiment results demonstrate that the GFRFT-based and PFLOM-FRFT-based methods can yield accurate estimation for the Doppler shift of aircraft scattering signal.更多还原