针对快速定点(FastICA)盲分离窄带干扰抑制算法对野点敏感和容易陷入鞍点或局部极值点的问题, 提出了一种基于独立成分分析(RobustICA)的盲分离窄带干扰抑制算法。该算法采用精确线性搜索与代价函数峭度相结合, 对峭度进行全局最优搜索, 增强了对野点和局部点的鲁棒性。仿真结果表明, 该算法抗强音频窄带干扰性能更优, 在信噪比为-14 dB~-11 dB时将分离矩阵成功收敛的概率提高了1个数量级, 增强了盲分离窄带干扰抑制算法的可用度。

A new interference suppression algorithm based on the Robust Independent Component Analysis(RobustICA) Blind Separation is proposed in Direct Sequence Spread Spectrum(DSSS) systems. In order to improve the robustness to the isolated points and local point, this algorithm combines the exact line search method with the kurtosis to get global optimum. Simulation results show that the proposed algorithm has a better performance for the multi-tone narrowband interference suppression by decreasing the probability of falling into the saddle point or the local point and enhancing the availability of the narrowband interference algorithm based on the blind source separation.