基于欠定盲分离的多目标微多普勒特征提取

An approach for extracting independent micro-Doppler characteristics of multiple targets based on underdetermined blind source separation

连续波雷达多目标回波中多种微多普勒特征分离问题采用独立成分分析方法实现,该方法在使用中存在较大局限性,要求待分离的微多普勒特征之间必须是统计独立的,且仅局限于恰定和超定的方程组求解问题。然而,在多目标雷达观测场景下,雷达接收的混叠回波的个数通常少于目标的个数,各目标的微多普勒特征可能存在相关性。为此,提出了一种基于欠定盲分离的多目标回波微多普勒特征分离方法。该方法可以从少数原始混叠回波中分离出多个目标的微多普勒特征,对待分离的微多普勒特征限制性弱。通过数值仿真,证实了该方法的可行性。

Abstract Nowadays the existing approach to extract various micro Doppler charac- teristics from backscattered signals by multi-targets usually employ blind source separation technique based on independent component analysis. However, this ap proach has a great limitation in real application. First, micro Doppler characteristics to be separated must be statistically independent; second, independent component analysis method is generally limited to the conditions of exact determined or over determined equations. Under the radar scene with multiple targets, the number of radar signal receivers is commonly less than the number of targets, as well as the micro Doppler characteristics may not satisfy the independent condition, which makes micro Doppler characteristics extraction more difficult. In order to solve these problems, an approach based on underdetermined blind source separation method ispro-posed in this paper. The approach not only work well for the situation that the number of received radar signals is less than that of targets, but also has fewer limi- tations in the micro Doppler characteristic to be separated. The simulation presented in this paper validates the proposed extraction approach.