基于ICA-R的复值信号抽取方法

Blind extraction of complex-valued signal using ICA-R

参考独立分量分析(independent component analysis with reference,ICA-R)通过引入参考信号而实现期望实值源信号的抽取,具有消除传统ICA输出顺序不确定性和显著降低运算量等优点.为此将ICA-R的优势拓展到期望复值源信号抽取.首先,将N维复值ICA问题转化为由其实部和虚部组成的2N维实值ICA问题;然后,利用期望源信号的实部参考信号或虚部参考信号进行ICA-R;最后,根据转换混合矩阵的结构特点,消除ICA-R抽取信号实部与虚部间的幅值不确定性,进而得到无附加相移的期望复值信号.计算机仿真和性能分析结果表明了所提方法的有效性.

Independent component analysis with reference （ICA-R） extracts only desired signals by incorporating prior information as reference signals. It has several advantages, such as eliminating the ambiguity of traditional ICA and significantly reducing computational load. ICA-R is extended to extract a complex-valued signal of interest. First, an N-dimension complex ICA is transformed into a 2N-dimension real ICA formed by a real part and an imaginary part. Then, ICA-R is applied to the real part or the imaginary part to give the corresponding parts of the desired signal, which are finally combined to form the estimated signal. By utilizing the characteristics of the transforming mixing matrix, the ambiguity between the real part and the imaginary part of the extracted signal is avoided, and the desired signal is finally obtained without phase error. Computer simulations and performance analysis demonstrate the efficacy of the proposed method.