Nonlinear algorithm for blind interference cancellation

Code division multiple access (CDMA) with direct sequence (DS) modulation is a promising multiple access scheme for wireless wideband mobile communications. In this paper, we introduce a simple nonlinear learning algorithm with hypothesis testing to perform blind interference cancellation with uncertainty of the training symbol timing. In addition, the training data set is very small. The method is successfully applied to a CDMA communication system, offering simple and efficient timing acquisition and symbol demodulation. We compare the method with a kurtosis optimization-based fast ICA method, conventional single user detector, and blind linear MMSE detector. It appears that our method is competitive with these and is even simpler than fast ICA.