Application of wavelet transform and empirical mode decomposition in the noise pre-processing of blind source separation

The blind source separation problem under noise is known as a hard problem. The performance of separation algorithm degrades with the decrease of SNR significantly. Wavelet transform and empirical mode decomposition are two typical analysis methods in time-frequency domain, especially for the processing of practical nonstationarity signals. In this paper, the principles of noise pre-processing for the two methods are discussed, and the performance is analyzed respectively. An algorithm that can be used in blind source separation of grading noise-pretreatment was proposed. Simulation results shows that pre-processing could make the present blind source separation work within the larger range of SNR and enhances the robustness of algorithm.