Independent component analysis for blind multiuser detections
Anthony Kuh, Xiaohong Gong · 2002
We apply a novel signal processing method based on independent component analysis (ICA) to blind multiuser receivers. ICA is well suited for blind multiuser detection problems as the criterion used to separate signals is a mutual information minimization principle which attempts to separate independent signals from mixed signals. When the cross-correlations between signature sequences are large, ICA has a better performance than decorrelating receivers and linear MMSE receivers.