Blind signal separation using oblique projection operators method

Recent decades, more and more people both from academic and commercial pay attention to blind signal separation (BSS). As an important part, independent component analysis (ICA) is a valid and effective solution to the problem of BSS, under the assumption conditions of ICA, a BSS algorithm using oblique projection operators is proposed in this paper. The autocorrelation matrix of mixing matrix is used to construct the objective function while the principle of maximum kurtosis is adopted to iterate and extract the component, and the mixing matrix can be obtained in a direct way. The description of the problem is demonstrated, and the detailed flow of the proposed method is listed. Simulation results show the suggested scheme is valid even when the weakest signal is less than -80dB to others.