Single-Channel Blind Source Separation of Communication Signals Using Pseudo-MIMO Observations

Abstract

To solve the problem of single-channel blind source separation of communication signals, this letter proposes a novel algorithm consisted of pseudo-multi-in multi-out (MIMO) observation construction and independent component analysis (ICA). A novel, effective way of constructing pseudo-MIMO mixtures for communication signals is proposed; sources are then retrieved by applying ICA on the artificial mixtures. The proposed scheme yields the following benefits: 1) efficiency in constructing pseudo-MIMO observations; 2) superior separation performance; 3) low computation complexity; and 4) low dependence on the prior information of sources. Experiments demonstrated edges of the proposed scheme over state-of-art algorithms in terms of both separation performance and computation complexity.