SUMMARYIn this paper, advanced interference cancellation strategies are considered and applied in DS-CDMA uplink receivers. Namely, we introduce a new successive interference cancellation (SIC) scheme in which blind source separation (BSS) techniques based on independent component analysis (ICA) are considered. Our main emphasis is on the oversaturated data in which situation easily arises for example in DS-CDMA systems when the number of active connections/users is high compared to number of chips per bit being used, that is, a highly loaded system. Proposed receiver structures combine the main benefits of pure BSS/ ICA and SIC methods: (i) inherent mitigation of various types of interference sources by BSS/ICA, (ii) robustness against parameter estimation errors due to BSS/ICA, (iii) greatly improved interference suppression capability due to novel combination of SIC ideology and ICA for oversaturated data. The last item in the list can also be seen as a way to somewhat circumvent the 'more sources than sensors'-problem, which is a challenging problem in BSS area. Numerical experiments with DS-CDMA uplink data are given to illustrate the achieved gains in capacity and bit-error-rate performance compared to conventional successive and parallel interference cancellation (PIC) schemes as well as more an advanced LMMSE-PIC receiver and variants of BSS/ICA techniques alone.