Nonlinear subspace blind adaptive multiuser detection in non-Gaussian channel using Kalman filter

Blind adaptive detection of spread spectrum signals in a DS-CDMA system in the presence of impulsive noise is considered. In the various detection schemes to mitigate multiple access interference (MAI), the blind adaptive approaches have advantages over non-blind ones with respect that they only need information of the desired user thus are more suitable for downlink implementation. Many previous detectors were based on the assumption that the channel noise was Gaussian. But many practical environments demonstrate non-Gaussian statistics that severely impair the performance of conventional linear detectors. In this paper a new detection structure is proposed that has a nonlinear pre-processing stage to mitigate the impulsive noise followed by a subspace Kalman filter to mitigate the MAI. Promising simulation results have been obtained which show that the new scheme has substantially improved robustness in impulsive noise channel than the linear detectors; while in the case of Gaussian noise channel, they are comparable.