Estimation and Elimination of Noise for ICA Based Blind Multiuser Detection

In this paper, two independent component analysis (ICA) based algorithms are proposed for blind multiuser detection (BMUD) in DS-CDMA systems. The first algorithm is Per-processing for Noisy ICA Based Blind Multiuser Detection that can reduce the noise in the detection system but it still has the accumulation of error. The second algorithm is Estimation and Elimination of Noise for ICA Based Blind Multiuser Detection which converses nonlinear noisy model into linear model to utilize some special characters of the new mixing matrix, and estimate the noise signal and eliminate the noise signal. This proposed approach can work well under low SNR conditions. The performances of the algorithms are evaluated using computer simulation. Simulations indicate improvements in terms of error probability and stability performance in detection.