基于二阶矩的雷达信号盲分离

针对雷达接收机在现代战场复杂电磁环境下接收到的混叠信号，提出了一种基于二阶矩的信号盲源分离方法。在混合信号球化过程中，对于具有加性白噪声的模型，构造了一组新的协方差矩阵，在信噪比不是很高的情况下，使其不会影响分离结果。在协方差矩阵对角化过程中，采用自然梯度的方法，避免分离矩阵更新过程中的求逆问题，提高了算法的实时性。仿真实验证明，在信噪比为-10 dB的条件下，对比FastICA算法，所提算法分离精度高，收敛速度快，为进一步的信号识别提供可靠依据。

Blind source separation of radar signals based on second order statistics

For mixed signals received by radar in the complex electromagnetic environment of modern battlefield,a blind source separation(BSS) algorithm is proposed which is based on second order statistics(SOS). In processing whiten mixed signals,a set of covariance matrices is constructed with a new method in order to reduce the influence of additive white noise when the signal-to-noise ratio(SNR) is low. In the process of covariance matrix group joint diagonalization,in order to avoid computating the converse matrix,the natural gradient method is used and the real-time of algorithm is improved. Computer simulation results prove that the separation precision of the proposed algorithm is higher than that of FastICA algorithm,and the convergence speed is also faster than that of FastICA algorithm,which provides reliable basis for further signal recognition.