一种并行主偏度分析算法及其在盲源分离上的应用

盲源分离是一种从混合信号中提取和恢复源信号的信号处理方法。在众多盲源分离算法中，主偏度分析算法是近年来出现的一种以三阶统计量为目标函数的盲源分离算法，其运算速度快于常规的盲源分离算法，但因其采用了串行的计算方式，在使用中存在误差累积问题。针对这一问题，本文在主偏度分析算法基础上进行改进，提出了一种并行主偏度分析算法。该算法以并行计算代替串行计算，可以同时估计出各个独立成分对应的方向，避免了误差累积问题。数值仿真实验表明，与主偏度分析算法相比，并行主偏度分析算法既保持了计算速度，同时提高了对源信号的估计准确性。

A Parallel Principal Skewness Analysis Algorithm and Its Application to Blind Source Separation

Blind source separation algorithm is utilized in extracting and recovering source signals from mixed signals. Among many different blind source separation algorithms, the principal skewness analysis algorithm is a blind source separation algorithm taking skewness as the objective function, which is faster than the conventional blind source separation algorithms. However, because of the serial calculation method, error accumulation exists in the calculation process. Based on the principal skewness analysis algorithm, this paper proposes an algorithm called parallel principal skewness analysis. This algorithm employs parallel orthogonalization instead of serial calculation, and can calculate the direction of each independent component simultaneously which avoids the error accumulation problem. Numerical simulation experiments show that the calculation accuracy of the parallel principal skewness analysis algorithm is improved compared with the principal skewness analysis algorithm, and the calculation speed is equivalent to it.