Spatial time-frequency distribution (STFD) utilizes the spatial time-frequency characteristics of chirp signals and effectively improves the direction of arrival (DOA) estimation performance. However, the existing methods based on the STFD matrix assumed that each source has good time-frequency point selection performance. In practice, the time-frequency point selection accuracy of chirp signals suffers from the signal-to-noise ratio (SNR). Especially in the case of low SNR, the time-frequency point selection error increased, leading to the reduction of the SNR and degradation of the DOA estimation performance in the time-frequency domain. To solve the above problems, we propose a time-frequency DOA estimation method based on multiple subarrays. The array is firstly divided into several overlapping subarrays, and the data received by the array is transformed from the element space into beamspace by using beamforming, which improves the source separation and the SNR. Then, chirp signals possess the ideal energy collection features in the time-frequency domain so that the time-frequency analysis tools are used for separated sources in beamspace to further improve the source separation and the SNR in the time-frequency domain. This results in better time-frequency point selection accuracy of chirp signals at a low SNR regime. Finally, the averaged STFD matrix can be obtained through averaging over multiple single-source time-frequency points in the time-frequency domain, and the DOAs can be obtained by combining with the subspace-based method. The theoretical analysis and simulation results indicate that compared with the existing STFD-based methods, the proposed method in this paper provides good performance on estimation and resolution in cases with low input SNRs due to beamspace processing. Furthermore, in cases where the DOAs between the coherent sources are closely spaced and the snapshot number is low, our proposed method significantly improves the performance of the DOA estimation.

空间时频分布（STFD）利用线性调频信号的空间时频特性，有效地提高了到达方向（DOA）估计性能。但是，基于STFD矩阵的现有方法假定每个源都具有良好的时频点选择性能。实际上，线性调频脉冲信号的时频点选择精度受到信噪比（SNR）的影响。特别是在低SNR的情况下，时频点选择误差会增加，从而导致SNR降低，并在时频域中降低DOA估计性能。为了解决上述问题，我们提出了一种基于多个子阵列的时频DOA估计方法。首先将数组分为几个重叠的子数组，利用波束成形将阵列接收的数据从单元空间转换为波束空间，提高了信源分离度和信噪比。然后，线性调频脉冲信号在时频域中具有理想的能量收集特征，因此时频分析工具可用于波束空间中的分离源，以进一步改善时频域中的源分离和SNR。这导致在低SNR体制下线性调频信号的时频点选择精度更高。最后，可以通过对时频域中的多个单源时频点求平均来获得平均STFD矩阵，并且可以结合基于子空间的方法来获得DOA。理论分析和仿真结果表明，与现有的基于STFD的方法相比，本文提出的方法在由于波束空间处理而导致输入SNR低的情况下，在估计和分辨率方面提供了良好的性能。此外，在相干源之间的DOA间隔很近且快照数量较少的情况下，我们提出的方法可显着提高DOA估计的性能。