Nonlinear blind interference cancellation with an approximate Newton algorithm for DS-CDMA system

Direct sequence (DS) code division multiple access (CDMA) is a promising technology for wireless environments with multiple simultaneous transmissions. However, the capacity of DS-CDMA systems is interference limited, especially when the multiple-access interference (MAI) is strong. The MAI can be cancelled blindly using minimum mean output energy (MMOE) adaptation. The conventional MMOE scheme has a linear structure and the performance is limited in most real life environments, which are nonlinear in nature. Therefore, a nonlinear MMOE scheme based on the Volterra expansion is proposed in this paper. In particular, the conventional Newton algorithm (RLS-recursive least square) is reformulated to fit the nonlinear case. An approximate Newton algorithm is developed to reduce the computational complexity and to improve the stability of the algorithm. A comparative performance analysis of the conventional matched filter, exact Newton algorithm and approximate Newton algorithm is carried out. The results show that the proposed approximate Newton method performs as good as an exact Newton method, both have great performance improvement compared to the conventional matched filter output.