Wavelet-FastICA-based separation method for single-channel and time-frequency overlapped signal in electromagnetic surveillance
Lihui Pang, Bin Tang · International Journal of Information and Communication Technology · 2017
In signal processing society, much attention has been paid in blind source separation (BSS) due to its 'blind' property and wide applications. However, there are still some open problems, such as single-channel BSS (SCBSS). In this paper, a SCBSS method called wavelet-fast independent component analysis (FastICA) is proposed for simultaneously received multi-system frequency-overlapped signals in a single-channel electromagnetic surveillance system. Firstly, wavelet is employed to decompose the single-channel recording into high dimensional data, as most of ICA algorithms rely on spatial (i.e., multichannel) analysis. Morlet wavelet is selected in this work for its non-orthogonality. Then, the method adopts FastICA to process the wavelet decomposition results, as so to find the independent components. Finally, the spectrum of the independent components (ICs) provided by FastICA is used to identify and recover the original sources. Numerical simulation results obtained in evaluating the proposed methodology's performance confirmed the effectiveness of the proposed algorithm and demonstrated its anti-noise superiority.