Blind Source Separation of Adjacent Group Targets Based on Coupling Scattering Center Removal

Abstract

Blind source separation (BSS) is an effective method to deal with the aliasing echo received by radar in the field of target detection. Single-channel BSS is a morbid problem of multi-channel BSS, but it has higher research value since it is more in line with practical application scenarios. Therefore, a single-channel BSS problem for adjacent group targets is studied in this paper. The coupling signals will be generated between targets when the distance between group targets is less than 10 times wavelength. It is difficult to accurately separate the source echoes of each target from the mixed echoes because the coupling echoes will have a certain influence on the mixed echoes of group targets. A new BSS method based on density clustering algorithm is proposed in this paper in order to solve this problem. Firstly, the strong scattering center points of group targets are obtained by radar imaging. Then, the strong scattering center points are clustered by the density-based clustering algorithm (DBCA). Next, the coupling echo is determined according to the variation of amplitude with azimuth, and the coupling signal is removed by the position information of the cluster. Finally, the mixed echo is separated, and the different separation methods are selected according to the spectrum of the mixed echo. 2-D imaging inverse operation is adopted if there is a single spectral peak; the empirical mode decomposition-fast independent component analysis (EMD-FastICA) algorithm is adopted if there are multiple spectral peaks. The simulation results show that this method can effectively separate the source signal from the echo containing coupling information within the error range.