【作者】 于红旗；

【导师】 张锡祥； 周一宇；

【作者基本信息】 国防科学技术大学 ， 信息与通信工程， 2007， 博士

【摘要】 宽带信号具有抗干扰能力强、分辨率高、携有的目标信息量大、与背景噪声相关性弱的特点,有利于目标检测、参量精确估计、目标特征提取,以及具有低截获和抗干扰等特性,已在雷达、声纳、通讯、地震勘探、射电天文、医学成像等众多领域得到了广泛的应用,如超宽带合成孔径雷达、无人机载合成孔径雷达、高距离分辨率雷达、宽带低截获概率雷达、超宽带通信等。在宽带情况下,由于阵列输出的相位差不仅仅依赖于DOA,同时还依赖于信号频率,频段的宽范围造成的相位差别无法忽略,在这种宽带情况下使用窄带阵列系统,将造成估计结果有较大的误差,甚至失效。因此,在无源侦察等应用中,必须针对宽带信号,研究有效的宽带信号阵列处理方法。同时,为满足对复杂调制样式、大时宽带宽积及大相对带宽等复杂信号侦察的需要,宽带信号阵列处理技术需要解决现有阵列处理技术适应大信号带宽能力弱、计算量大、无法在现有装备上应用等问题。本文结合国家自然科学基金项目,针对以上问题,研究了基于角度预估的聚焦矩阵构造方法、不需要角度预估的聚焦矩阵构造方法、子空间投影正交性测试算法、宽带信号DOA估计快速算法以及波束域宽带信号DOA估计方法。第二章研究了信号带宽对窄带源个数估计方法及MUSIC算法的影响,总结了宽带信号阵列到达角估计的信号模型,提出了几种宽带阵列接收信号的模型。第三章研究了基于角度预估的宽带信号DOA估计方法。提出了基于信号子空间分解的宽带相干信号子空间DOA估计方法,证明了该方法与TCT方法具有相同的性能,但运算量小于TCT方法;提出了基于各频点阵列流形标准正交基的聚焦矩阵构造方法,通过仿真验证了基于该聚焦矩阵的宽带信号DOA估计算法性能;。第四章研究了不需要角度预估的宽带信号DOA估计方法。研究了基于Bessel函数的宽带相干信号子空间DOA估计方法,通过对阵列流形展开以分离频率和角度变元,在构造聚焦矩阵时仅仅使用与频率相关的部分,避免了对预估角度的需求;提出了基于三次样条插值的宽带相干信号子空间DOA估计方法,根据频率对阵列采样数据进行插值构造不同间距虚拟阵列采样,进而获得DOA估计。对一致聚焦算法进行改进,提出了基于奇异值分解的改进一致聚焦算法。第五章研究了投影子空间正交性测试方法,提出了频域子空间投影正交性测试方法,该方法根据构造的搜索向量与每个频点的噪声子空间的正交性得到最终的空间谱,避免了TOPS方法中对参考频率处信号子空间的估计,降低了运算复杂度,提高了算法性能。第六章研究了宽带信号DOA估计快速方法,将多级维纳滤波、传播算子方法、快速子空间分解方法、Householder QR分解方法等引入到聚焦矩阵构造过程中,根据每种快速算法不同的特点提出了相应的聚焦矩阵构造方法,所得到的算法大大降低了运算量,且采用乘加形式,便于硬件的实时实现。第七章研究了波束域宽带信号DOA估计方法,总结了频率不变波束形成在宽带信号DOA估计中的应用,提出了两种设计频率不变波束形成的方法,并将其应用到宽带信号DOA估计中。更多还原

【Abstract】 Wideband signals possess the characteristics of high resolution, large information quantity, weak correlation with background noise, low probability of intercept and anti-interference, which benefit target detection, precision parameter estimation and target feature extraction. Therefore wideband signals are widely used in radar, sonar, communication, seismic exploration, radio astronomy, medical imaging and other fields, such as ultra-wideband synthetic aperture radar, UAV synthetic aperture radar, high range resolution (HRR) radar, wideband LPI radar and ultra-wideband communication.For wideband signals, the phase difference between sensor outputs is no longer just dependent on the DOA alone, but also depends on the temporal frequency, which has a wide range. The difference brings a large deviation, or even failure, to the narrowband DOA estimation algorithms from the true parameters. Therefore, in passive reconnaissance, effective wideband array processing methods should be proposed. Meanwhile, in order to meet the need of reconnoitering signals of complex modulation types or of large time-bandwidth product, wideband array processing techniques should cope with the problems of existing array processing techniques, such as the weak adaptation to large signal bandwidth, tremendous amount of computation and incompatibility with existing equipments.Supported by the Natural Science Fund of China, this dissertation carries on the following research into the problems above: the formation of focusing matrix with and without preliminary DOA estimation, test of orthogonality of subspace, fast wideband DOA estimation algorithms and beam-space wideband DOA estimation algorithms.Some basic problems are discussed in Chapter 2. The influence of bandwidth on narrowband sources’ number detection and MUSIC algorithm is studied. Some ways of modeling the signals received by wideband array are developed.Wideband DOA estimation algorithms based on preliminary DOA estimation are studied in Chapter 3. A new wideband CSSM algorithm based on signal subspace decomposition is developed, the new algorithm has the same performance as that of TCT algorithm, but has less computation burden. A new method of forming focusing matrix, basing on SVD of the array manifold, is proposed. In addition to that, a new ISSM method, which bases on the synthesis of preliminary DOA estimation and ISSM method, is developed through the replacement of array manifold.Wideband DOA estimation algorithms without preliminary DOA estimation are studied in Chapter 4. The CSSM method based on Bessel function expansion is studied; the frequency and DOA are separated through the expansion of the array manifold, the part related to frequency is used to form the focusing matrix. A new CSSM method based on cubic spline interpolation is developed, the virtual array outputs are formed by cubic spline interpolation and are allied at the reference frequency, thus narrowband DOA estimation can be used. The consistent focusing method is improved, in which unitary matrix constrain is added to the focusing matrix to gain better performance.The algorithm of subspace’s orthogonality testing is studied in Chapter 5. A new algorithm is proposed basing on the frequency domain subspace’s orthogonality testing. The new algorithm avoids the estimation of signal subspace at the reference frequency and has better performance and lower computation burden compared with TOPS.The fast algorithms in wideband DOA estimation are studied in Chapter 6. The multi-stage Wiener filter, propagator method, fast signal subspace decomposition and Householder QR decomposition are introduced into wideband DOA estimation. The proper focusing matrices are proposed according to the characteristics of the above fast algorithms. The proposed methods are in the form of multiply-accumulator, so they are real-time realizable in hardware.The wideband beam-space DOA estimation algorithms are studied in Chapter 7. The narrowband beam-space DOA estimation algorithms are summarized and developed to wideband scenario. The wideband DOA estimation based on frequency invariant beamforming is summarized. Two new methods that construct frequency invariant beamforming are developed. These methods are used in the wideband DOA estimation, and the performance is verified through simulations.更多还原