波达方向（Direction of Arrival, DOA）估计作为阵列信号处理的一个重要研究方向，在雷达、通信等领域有着广泛的应用前景，目前，它已成为阵列天线系统和智能天线的关键技术。
波达方向估计的主要研究内容之一是如何从背景噪声中准确地估计信号的方位（而这些背景噪声各种各样，大致从简单的高斯空间白噪声环境到复杂的未知空间色噪声环境）。实际工程中，最常出现的是复杂噪声环境下的波达方向估计问题，由此，不同噪声环境下的DOA估计算法研究，不仅有着重要的理论意义，也具备较高的实际应用价值。
线性调频（Linear Frequency Modulation ,LFM）信号是非平稳信号中普遍存在的一种信号形式，由于它的非平稳特性，其参数估计较一般的信号复杂，得到了学者们的广泛关注。本文主要就不同噪声环境下的宽带LFM信号波达方向估计进行研究。
论文首先介绍了信号模型与阵列形式及其传统的时频子空间DOA估计算法，然后针对不同的加性噪声环境，本文着重研究了如何在不同噪声环境下得到准确的、高分辨的波达方向。
本文的主要工作和贡献有：
1、研究了宽带LFM信号DOA估计的几种算法，重点介绍了自适应信号稀疏分解方法——MP（Matching Pursuit）算法。计算机仿真实验表明，在高斯白噪声环境下，对比于传统的时频信号子空间DOA估计算法， MP算法性能较好，尤其在低信噪比情况下仍然有较高的估计精度。
2、给出了 冲击噪声环境下的高分辨DOA估计算法。首先，给出了基于无穷范数及Screened Ratio原理的MP（简称“IN-SR-MP”）波达方向估计算法，即利用无穷范数归一化先进行预处理，接着通过Screened Ratio原理的启发来构建MP分解过程中进行匹配所需的相关矩阵，最后利用传统MP算法，实现LFM信号波达角估计；其次，鉴于MP算法存在计算量大的问题，又推导出基于无穷范数归一化预处理的传播算子（简称“IN-PM”）波达方向估计快速算法。上述两种算法的估计性能均优于常规的FLOM-TF-MUSIC算法、FLOM-MP算法。
3、在高斯色噪声环境下，详细介绍了基于线性时频变换（FRFT）的四阶累积量MUSIC算法对非平稳宽带LFM信号的DOA估计过程，针对上述算法所存在的低信噪比下估计性能失效及运算量大的问题，又给出了基于FRFT的传播算子的DOA估计算法。仿真实验验证，基于FRFT的四阶累积量MUSIC算法能够用来处理高斯色噪声。同时，实验亦表明FRFT-PM算法估计性能略优于FRFT-CUM-MUSIC算法，和MP算法性能相当。
4、在两种噪声共存的混合加性噪声环境下，设计了基于FRFT的传播算子新算法，并将基于FRFT的传播算子算法及其新算法运用到混合加性噪声环境下，仿真实现了混合加性噪声环境下的LFM信号DOA估计。

Direction of arrival (DOA) estimation is one of the most important research areas of array signal processing which has found wide applications in a variety of fields ranging from radars, communication. Especially, it plays an essential role in the passive detection of emerging array radar and smart antenna systerm. One of the main studies for algorithms of DOA estimation is how to accurately estimate signals’ orientation under the background noise, while the background noise is various, ranging from the simple white Gaussian noise circumstance to the complex unknown space colored noise circumstance. In practice application, one always encounters the problems of DOA estimation in complex noise. Therefore, the research concerned with the DOA estimation methods in various noise cases has the important theory significance and the practical application value.
Wideband linear frequency modulation (LFM) signals exist widely in nonstationary signals whose frequency contents evolve with time. As the nonstationary feature of LFM signals, parameters are harder to detect than stable signals, and parameter estimation of LFM signals is studied all over the world. This dissertation mainly involves direction-of-arrival estimation for wideband signals in various and complex noise fields.
Firstly, the thesis introduces some kinds of traditional algorithms, signal models and array forms, and does analysis and comparison on some classical algorithms. Then, it gives main research on how to get more accurate direction of arrival for wideband signals in various noise fields.
The following is the summarization of the main work:
(1) The methods of direction-of-arrival estimation for wideband LFM signals are first investigated, and adaptive signal MP decomposition algorithm is discussed. Compared with the traditional DOA estimation algorithms based on the subspace method under time-frequency analysis in the white Gaussian noise case, numerical simulations of the resolution show that the MP decomposition algorithm’s performance is superior to the conventional STFD based methods, especially in the case of lower SNR.
(2) The high-resolution array direction finding algorithms are deduced in the presence of impulsive noise. Using infinite-norm(IN) normalization pre-processing, then using the principle of Screened Ratio to construct an adaptive over-complete matrix dictionary, the infinite-norm normalization pre-processing based MP（IN-SR-MP）and Propagater Method（IN-PM）alterative algorithms are developed, whose performances are superior to the general FLOM-TF-MUSIC and FLOM- MP algorithms, and IN-PM algorithm is simpler and speed is faster.
(3) Under Gaussian colored noise background, the essay introduces the process of the DOA estimation for LFM signals minutely, using the methods of the 4th-order cumulant MUSIC based on linear frequency transform (FRFT). At the same time, for the existence problem of the low estimated accurancy resolution and larger computation of the above algorithm, one new improved method for DOA estimation is proposed based on FRFT. Computer simulations show that while FRFT-CUM-MUSIC algorithm can be used in Gaussian colored noise background, the resolution of the proposed algorithm is better than the above algorithm and equal to MP algorithm.
(4) A new improved method for DOA estimation based on FRFT-PM in complex and additive noise is proposed, which includes two different noises. Numerical simulations of the resolution show that the new improved algorithm’s performance is superior to the previous algorithm, and they can be also used to estimate DOA of LFM in complex and additive noise.