Abstract

Single channel blind source separation is an appealing and challenging task in blind source separation. In this paper, we develop a robust technique for single channel blind source separation by combining slope ensemble empirical mode decomposition (S-EEMD) and independent component analysis (ICA). The proposed S-EEMD algorithm is firstly used to decompose mixed signal's fluctuation and trend of different scales step by step and generates a series of data sequence which called intrinsic mode function (IMF). After gaining a set of IMFs, we then adopt ICA algorithm to select the independent components which are similar to source signals and recover source signals. Validation of the proposed method is performed with extensive experiments on toy and real-life datasets respectively. Results demonstrate that compared with the EEMD method, our method has lower computational complexity and better robustness. Furthermore, our method can effectively overcome edge effect in EEMD algorithm.

Paper Details