This paper presents a novel mixing matrix estimation method based on a frame cluster analysis for application to the dynamic system of radar signals under an underdetermined blind source separation. The received signals are first processed using an adaptive denoising method. They are then divided into different appropriate length frames. Next, the frame type is determined and each frame is processed in accordance with its type. Short-time Fourier transform is used to transform the mixed signals from time domain to time–frequency domain. A transformation matrix is introduced to resolve the issue of the traditional clustering algorithm not being applicable within the complex field. After introducing the transformation matrix, an innovative single-source point detection algorithm and an estimation algorithm for the number of source signals are proposed. Finally, the mixing matrix is estimated via a fuzzy c-means clustering algorithm based on the characteristics of complex numbers. The simulation results show that the proposed algorithm improves the estimation accuracy of the mixing matrix considerably. Further, it has evident advantages for blind estimation of the mixing matrix for time-varying radar signals in the dynamic system.

中文翻译：

---

UBSS模型下动态系统中时变雷达信号的混合矩阵估计算法

本文提出了一种新的基于帧簇分析的混合矩阵估计方法，适用于欠定盲源分离下的雷达信号动态系统。首先使用自适应降噪方法处理接收到的信号。然后将它们分成不同的适当长度的帧。接下来，确定帧类型并根据其类型处理每个帧。短时傅里叶变换用于将混合信号从时域变换到时频域。引入变换矩阵来解决传统聚类算法在复杂领域不适用的问题。引入变换矩阵后，提出了一种创新的单源点检测算法和源信号个数估计算法。最后，根据复数的特点，通过模糊c-means聚类算法估计混合矩阵。仿真结果表明，该算法显着提高了混合矩阵的估计精度。此外，它对于动态系统中时变雷达信号的混合矩阵的盲估计具有明显的优势。