A new approach to soft parallel interference cancellation for synchronous CDMA

An attribute of the decorrelating detector (DD) is the ease in which the a posteriori bit probabilities may be computed. In this paper, we use these a posteriori bit probabilities (e.g.. π_i ≐ P(b_i = +1 | y_)) to replace the standard a priori probabilities (e.g. π_i ≐ P(b_i = +1) ≡ 1/2 ) and find soft bit estimates that optimize particular cost functions under the Bayesian rule. The soft multiple access interference (MAI) bit estimates,are suitable for a parallel interference canceller (PIC) and create a new genre of CDMA detection algorithms. The bit error rate (BER) performance is shown to better approximate that of the optimal MAP detector than other soft PIC methods. The additive white Gaussian noise assumption (required for the interference canceller) is validated by the use of an adaptive beamformer.