Effect of channel estimation and multipath on interference cancellation employing blind source separation in the DS-CDMA uplink

In this paper we consider DS-CDMA uplink interference cancellation based on the blind source separation (BSS) technique. While BSS assisted interference suppression compared to linear receivers is quite well reported, a comparison of BSS with nonlinear receivers is overlooked in the literature. Huovinen (2003) was the first one to make this comparison in the AWGN channel with perfect channel information. In this paper we make further evaluation. In particular, the effects of multipaths, related channel estimation errors and additional external interference (like adjacent channel interference or intentional jamming) are examined and compared to conventional successive interference cancellation (SIC) and M-stage parallel interference cancellation (PIC) schemes. We demonstrate by numerical experiments that benefits of using BSS exists. Moreover, results indicate that the performance gain compared to SIC and PIC is even greater in the multipath scenario and in the presence of channel estimation imperfections.