时分多输入多输出（Time Division Multiplexing-Multiple Input Multiple Output，TDM-MIMO）毫米波雷达因其简单的硬件结构而被广泛使用，但时分发射带来了目标角度速度耦合问题，并降低了雷达的目标分辨能力。如何避免参数耦合对估计精度的影响，改善时分体制雷达的目标分辨能力，是当前研究的重点。本文主要围绕TDM-MIMO雷达观测运动目标和微动目标场景，开展关于雷达阵列波形设计和目标参数估计方法的研究。本文主要工作如下：针对TDM-MIMO雷达的观测运动目标时的速度角度耦合问题，提出了一种基于循环嵌套结构的雷达阵列波形设计方法和一种基于高维图的参数估计方法。前者通过将均匀的接收阵列循环地嵌入稀疏的MIMO等效阵列，将连续的脉冲波形循环地嵌入各天线发射波形中，保证了发射天线切换和TDM发射循环间目标回波相位变化的均匀性。在此基础上，高维图在传统二维角度速度图的基础上增加了用于描述上述相位变化参数的新维度，从而避免了非均匀采样带来的旁瓣问题，实现了对多目标的检测和角度速度估计。仿真和实测结果表明，所提出的方法能够解决角度速度耦合问题，且其精度能达到克拉美罗界（Cramer-Rao Bound，CRB）。针对TDM-MIMO雷达的分辨能力问题，提出了一种基于波形分集的联合参数估计方法。在循环嵌套结构的基础上，通过在目标探测中使用多种不同的波形配置，可以获得多个不同的模糊函数下的目标探测结果。融合这些结果可以减少时分体制导致的模糊问题，使TDM-MIMO雷达获得和同等孔径的单发多收雷达相同的分辨能力。实验结果表明，所提出的方法可以提升雷达系统的分辨能力，且适用于多种复用体制下的多雷达或连续观测场景。针对TDM-MIMO雷达探测微动目标问题，提出了一种基于微动回波谐波模型的速度、角度和微动频率估计方法。本文将波形未知的周期性微动的回波建模为一系列谐波，然后基于最大似然原理设计了基于高维图的微动参数估计方法。本文进一步通过对谐波模型的正则化，推导了微动参数估计的CRB，实现了任意微动波形下对最大似然估计性能的定量描述。仿真和实测实验结果表明，所提出的建模和估计方法能解决实际中的目标微动频率估计问题，且精度与推导的渐近性能契合。
The use of Time Division Multiplexing Multiple Input Multiple Output （TDM-MIMO） millimeter wave radar is widespread due to its simple hardware structure. However, time-division transmission brings about the problem of target angle velocity coupling and reduces the radar‘s resolution. How to avoid the impact of parameter coupling on estimation accuracy and improve the target resolution of time-division radar is currently the focus of radar research. This paper focuses on researching radar array waveform design and target parameter estimation methods for TDM-MIMO radar and moving targets. The main work of this article includes:To address the velocity angle coupling problem when observing moving targets with TDM-MIMO radar, this paper proposes a radar array and waveform design method based on cyclic nested structure and a parameter estimation method based on high-dimensional map. The former ensures the uniformity of target echo phase changes between transmission antenna switching and TDM transmission loops by cyclically embedding a uniform receiving array into a sparse MIMO equivalent array and cyclically embedding continuous pulse waveforms into each antenna’s transmission waveform. Based on it, the high-dimensional map with new dimensions to describe the phase change parameters mentioned above, compared with the traditional two-dimensional angular velocity map, is proposed to avoid the sidelobe problem caused by non-uniform sampling and thereby detect multiple targets and estimate their angle and velocity. The simulation and experimental results show that the proposed method can solve the problem of angular velocity coupling, and its estimate accuracy reaches the Cramer-Rao Lower Bound （CRLB）.A joint parameter estimation method based on waveform diversity is proposed to improve the resolution of TDM-MIMO radar. On the basis of the cyclic nested structure, multiple different waveform configurations can be used in target detection to obtain target detection results under different ambiguity functions. Fusing these results can reduce the ambiguity caused by the time division system, enabling TDM-MIMO radar to achieve the same resolution as Multiple Input Single Output （MISO） radar with the same aperture. The experimental results show that the proposed method can improve the resolution of TDM radar systems and is suitable for multiple radar or continuous observation with different multiplexing methods.A method for estimating the micro motion frequency of micro motion targets detected by TDM-MIMO radar is proposed. The method models the periodic micro motion echoes with unknown waveforms as a series of harmonics, and is designed based on high-dimensional map according to the maximum likelihood principle. This paper further derives the CRLB for micro motion parameter estimation by regularizing the harmonic model, achieving a quantitative description of maximum likelihood estimation performance with arbitrary given micro motion waveform. The simulation and experimental results show that the proposed modeling and estimation method can solve the problem of target micro motion frequency estimation in practice, and the accuracy is consistent with the derived asymptotic estimation error.
