Single-channel Blind Source Separation Algorithm Based on Water Area Noise Characteristics

Abstract

Aiming at the problem that the characteristics of the water area noise in the underwater acoustic communication have a great impact on the performance of the single channel blind signal separation algorithm, a blind source separation algorithm based on noise characteristics of fixed waters is proposed. In this paper, a theoretical model of fixed water noise is established, and the statistical characteristics of noise are obtained by experimental test. Then, according to the obtained noise characteristics, a series of intrinsic mode functions obtained by variational mode decomposition are sparse processed to reduce the interference of noise components, and the signals are reconstructed to form virtual channel signals. Finally, the Fast-ICA algorithm is used to realize the blind source separation of the signals. LFM, sine wave and sawtooth wave are used as transmission signals for experimental verification. The results show that compared with the direct VMD + Fast-ICA algorithm, the correlation between the transmitted signals and the separated signals obtained by the algorithm proposed in this paper is higher, and the difference of correlation coefficient between different separated signal components is larger, that is, it has better signal separation performance and feature retention performance.