Comparison of nonlinear interference and blind source separation techniques in the DS-CDMA uplink

In this paper we compare the performance of conventional nonlinear interference cancellation and blind source separation (BSS) techniques. Namely, serial interference cancellation (SIC). M-stage parallel interference cancellation (PIC) and BSS based on independent component analysis (ICA) are employed in the uplink of direct-sequence code division multiple access (DS-CDMA) system. Existing studies on BSS/ICA in DS- CDMA have mainly concerned blind interference cancellation, which have been argued by the ability of BSS to deal with unknown (interfering) signals while demodulating a specific user. The performance of BSS/ICA techniques compared to conventional single user detection and blind (linear) multiuser detectors are quite well understood. In this paper we highlight the effectiveness of BSS/ICA against conventional nonlinear interference cancellation schemes. The systems with and without additional external interference like adjacent channel interference or jamming, are considered. Numerical experiments are given to evaluate the performance in different settings. They indicate clear performance gains in comparison to SIC and PIC with just a few stages when no jamming is present. However, even with a moderate level of additional jamming, BSS/ICA can outperform both the nonlinear schemes. The results thus show how the blindness of BSS/ICA makes it a robust method in the presence of intentional/unintentional out-of-cell interference.