基于单个频点的水声信号盲源分离

Blind Source Separation of Underwater Signals Using Single Frequency Bin

该文提出基于单个频点的卷积信号盲源分离方法,利用该方法不但可以有效克服频域盲分离过程中排序 不确定问题,而且在分离过程中,无需考虑幅度不一致问题。将该方法用于水声信号的盲分离,仿真结果表明基 于单个频点盲源分离方法能够很好地分离水声卷积混合信号。与基于两个频点盲源分离方法相比较,其分离效果 更优,并且能有效节省CPU运算时间,因而更适合于对信号进行实时处理。

A new blind signal separation method, where only one frequency bin is used, is presented to separate the underwater signals. Using this method, the indeterminacy in amplitude and permutation can be eliminated, and the validity of this method is confirmed by simulation experiment. Compared with the separation algorithm based on two frequency bins, this new method has better separation performance and less operation time. Therefore, the separation algorithm based on single frequency bin is more suitable for real-time signal separation.