【作者】 杨涛；

【导师】 苏涛；

【作者基本信息】 西安电子科技大学 ， 信号与信息处理， 2014， 博士

【摘要】 多输入多输出(Multi-Input Multi-Output, MIMO)雷达是目前雷达技术领域的研究热点，其特点是每个发射天线可以独立发射不同的波形，与相控阵雷达所有阵元发射相同的波形相比，MIMO雷达的波形分集能力带来更多的发射自由度。可以通过设计发射波形灵活地设计期望发射方向图的形状。同时，MIMO雷达的波形分集能力是提高雷达资源利用率、实现雷达资源自适应分配调度的重要手段。本论文结合具体科研任务，主要围绕窄/宽带MIMO雷达的波形设计、稳健自适应波束形成方法和新型雷达实时信号处理系统等问题展开研究。论文的内容可概括为如下四个部分：第一部分研究了窄带MIMO雷达发射波形设计方法。针对窄带MIMO雷达在给定的期望发射波束方向图情况下，如何在线快速设计具有良好的自相关和互相关性能的发射波形问题，提出了基于协方差矩阵空时构造的MIMO雷达发射波形设计方法。首先将发射波形矩阵分解为空域方向矩阵和时域信号矩阵。然后通过一组发射导向矢量构造空域方向矩阵，通过利用MUSIC方法获得这组发射导向矢量的指向角度，而时域信号矩阵是由(0,1)矩阵和离线设计的正交波形集构造获得。实验结果表明该方法避免了复杂的迭代和优化过程，与已有方法相比，所提方法具有相近的发射方向图综合性能，但其计算复杂度明显降低。第二部分研究了宽带MIMO雷达发射波形设计方法。首先针对一维均匀线阵宽带MIMO雷达，提出了一种一维宽带MIMO雷达频率不变发射方向图快速综合方法，该方法首先利用均匀线阵的发射方向图与发射波形的频谱具有一维傅立叶变换的关系，去除了发射方向图与发射波形频率的相关性；然后建立了具有恒模约束条件的时域波形最优匹配频率响应的代价函数；最后采用交替优化的思想求解出恒模发射波形。为了匹配二维宽带MIMO雷达系统给定的期望发射方向图，将一维均匀线阵情况下提出的方法进行扩展，通过空域、频域变量的代换及二维逆傅立叶变换去除了发射方向图与波形频率的相关性。并且通过交替优化求解得到具有恒模性质的时域宽带发射波形。提出的方法能够有效地改善宽带MIMO雷达发射波束方向图频率色散问题，并且避免了多维优化问题中复杂的多维搜索，明显降低了运算复杂度，仿真实验验证了所提方法的有效性。第三部分研究了稳健自适应波束形成方法。针对导向矢量失配和样本协方差矩阵估计误差导致自适应波束形成性能下降的问题，提出一种基于波束域导向矢量估计的稳健方法。该方法首先利用期望信号角度失配区域的补集构造波束域转换矩阵，克服了当样本协方差矩阵中存在期望信号分量时现有算法出现有用信号对消的缺点。其次推导出波束域导向矢量估计方法，并将其转化为包含恒模约束的非凸二次约束二次规划问题，然后采用半正定松弛规划估计实际的波束域导向矢量。仿真验证了所提方法的有效性。第四部分研究了新型雷达实时信号处理系统。针对新型雷达系统（如MIMO雷达，认知雷达等）的结构特点，首先对雷达实时信号处理系统的体系结构、通用性设计等关键技术进行了研究。提出了一种基于交换的开放式雷达系统架构，该系统架构简化了复杂的异构多模块互连方式，有助于提高任务分配、软件设计和调试的效率，并适用于不同需求的雷达系统。然后在此系统架构下，设计了一种新型通用雷达实时信号处理机，该信号处理机采用模块化设计，具有处理能力强，维护扩展性好，可靠性高等特点。最后结合MIMO数字阵雷达系统的具体应用，设计了核心算法节点的任务分配方案，包括窄带MIMO接收处理、宽带波束形成和ISAR成像，并且分析了任务分配方案的实时性能。由于设计的新型通用雷达实时信号处理机具有良好的体系架构，通过配置不同的处理功能模块可以应用于不同的实时信号处理场合。更多还原

【Abstract】 Multi-input-multi-output (MIMO) radar has been intensively researched recently.MIMO radar is defined as a radar system with multiple antennas to simultaneouslytransmit diverse waveforms freely, as well as multiple antennas to receive the radarreflected signals. Compared with conventional phased-array radar, the waveformdiversity makes MIMO radar have greater flexibility to achieve the desired transmitbeampattern. Meanwhile the waveform diversity of MIMO radar is an important meansof improving resource utilization of radar system and implementing adaptive allocationof radar resources. Combining with a research project, the work of this dissertation isfocused on narrowband/wideband MIMO radar waveform design, robust adaptivebeamforming and new type radar universal real-time signal processing system.The main contents of the dissertation are summarized as follows.The first part focuses on narrowband MIMO radar waveform design. In order tosynthesize a desired spatial beampattern, which is mapped into a waveform correlationmatrix, a computationally attractive method is proposed. This method decomposes thewaveform matrix into a spatial direction matrix (SDM) and a temporal division matrix(TDM). Then the SDM is formed by using a set of transmit steering vectors, whoseangles can be estimated by MUSIC, and the TDM is constructed by (0,1)-matrix andorthogonal (uncorrelated) waveforms. The proposed algorithm leads toconstant-modulus transmit signals with good auto and cross-correlation property.Furthermore, this algorithm achieves almost the same performance as the cyclicalgorithm while having lower computational complexity than the existing methods. Anumber of numerical examples are presented to demonstrate the effectiveness of theproposed algorithm.The second part focuses on wideband MIMO radar waveform design. For onedimension (1-D) MIMO radar, a fast wideband transmit beampattern synthesisalgorithm based on frequency invariant beamformer (FIB) is proposed, the spectrum ofwideband MIMO radar transmit waveforms are computed by FIB, then alternatingmatrix fitting method is used to design unimodular sequences. For matching the desiredtransmit beampattern of two dimension (2-D) wideband MIMO radar system, theproposed method in1-D wideband MIMO radar is expanded to2-D wideband MIMOradar. In order to eliminate any dependency on frequency of emiting waveform,2-Dfourier relationship between uniform rectangular array (URA) transmit beampattern and spectrum of emiting waveform is used. Then cost function of waveform matchingfrequency response is established, which imposes a constraint on constant modulus.Finally, using alternating matrix fitting method to design unimodular sequences. Theproposed method effectively mitigates frequency dispersion of wideband MIMO radartransmit beampattern. The proposed method requires no matrix inversion and has muchlower computation complexity compared with wideband beampattern formation viaiterative techniques (WBFIT) due to the significant reduction of iteration order. Thenumerical simulations have proved the validity of the proposed algorithm.The third part focuses on robust adaptive beamforming. In order to solve theproblem of performance degradation due to the imprecise knowledge of the arraysteering vector and inaccurate estimation of the sample covariance matrix, A newapproach based on beamspace steering vector estimation for robust adaptivebeamforming is presented. Firstly, by using the complementary set of the spatial sectorin which the actual steering vector lies, beamspace transformation matrix can beconstructed to ensure that the signal of interest is removed from the samplingcovariance matrix. Then a method for beamspace steering vector estimation is derived,and mathematically expressed as the nonconvex Quadratically Constrained QuadraticPrograms (QCQP) problem with one non-convex quadratic equality constraint, whichcan be successfully solved by using SemiDefinite Relaxation (SDR) techniques.Simulation results show the effectiveness of the proposed algorithm．The fourth part is contributed to new type universal radar real-time singal processor.Aming at the structure characteristics of new radar systems (such as MIMO radar,Cognitive radar, etc), the key techniques of radar real-time processing system such assystem architecture and universal design are investigated. An open system architecturefor radar based on switch network is proposed. This system architecture simplifies thecomplex of multiple mudules interconnect, which contribute to improve the efficiencyof task allocation, software design and debugging. Furthermore, this system architectureis suitable for different requirements of radar systems. Based on the proposed openradar system architecture, a new type universal radar real-time signal processor,consisting of fiber processing modules, universal processing modules and high densitystorage module, is designed. This signal processor adopts the concept of modulizeddesign, with high processing proformance, good scalability and high reliability. ForMIMO digital array radar, MIMO receive proceesing method, wideband beamformingand ISAR image are important issues, thus the task allocation scheme and analysis ofreal-time requirement are investigated. The reconfigurable feature of the new universal radar real-time signal processor gives rise to reusability of hardware and scabability.The designed new universal radar real-time signal processor can be comprised ofdifferent modulars for various radar systems, consequently, reusability of hardwaresupporting multiple radar systems, including radar reconnaissance receiver and echosimulator for radar, etc.更多还原