Blind separation of instantaneous linear mixtures of cyclostationary signals
Rongbo Wang, Chaohuan Hou, Dongliang Chen · 2010
Blind separation of signals, which consists of recovering signals only from observed instantaneous linear mixtures without a priori information about the mixing matrix, is an important problem in many practical applications including radar, sonar, wireless communication and so on. For blind separation of cyclostationary signals, proposed a novel approach (PM) which dropped a pre-whitening stage usually required by many second-order statistics based methods such as AMUSE and SOBI algorithms. However, the PM algorithm has a fatal disadvantage in that it cannot fully separate all signals in the circumstance of more than two sources. To overcome this drawback, we proposed an improved PM algorithm in this paper. Several numerical simulations are also provided to illustrate the effectiveness of the proposed method.