Linear parallel interference cancellation using fixed weighting factors for long-code CDMA

Abstract

Linear weighted multistage parallel interference cancellation (PIC) implements exactly the family of polynomial expansion detectors. For long-code CDMA, a set of optimal weights is found which minimizes the ensemble averaged mean squared error (MSE) over random codes. The weights are dependent on moments of the eigenvalues of the correlation matrix, where exact expressions are derived. The loss incurred by averaging rather than using the optimal, time-varying weights is practically negligible.