Extraction of a weak co-channel interfering communication signal using complex Independent Component Analysis

Independent Component Analysis (ICA) algorithms taking advantage of the potential non-circular property of complex signals have been recently derived and shown to lead to improved performances. We investigate the performance of three ICA approaches to extract a weak co-channel interfering communications signal from a television broadcast signal over varied interference-to-noise ratios: complex maximization of non-Gaussianity by Novey et al., RobustICA by Zarzoso et al., and complex fixed-point algorithm by Douglas. Findings show improved performance and sensitivity to the prewhitening step as the number of sensors increases.