摘要

In order to improve the communication quality in the extremely-low-frequency communication effectively, a kind of new insight based on analog circuits and linear filtering method for interference suppression is proposed, the latter mainly achieves by the joint diagonalization of the correlation matrices. The proposed algorithm takes place the filter coefficients by a linear combination of eigenvectors, the problem of interference suppression is converted into the solution of constructed coordinates. Besides, three kinds of filters are designed based on the mean square error criterion, the derivation process is given in detail, and several performance evaluation indices are also provided. And the derivation results show that the proposed algorithm can become common filters in speech enhancement when the number of chosen eigenvectors is a certain value. Finally, some experiments are performed in a laboratory environment. And the experimental results show that the proposed algorithm can suppress the interference effectively, and the compromise between the number of interference suppression and the degree of signal distortion can be controlled by changing the number of chosen eigenvectors, the proposed algorithm has potential performance gain.