Channel Estimation and Multiuser Detection for Frequency - Nonselective Fading Synchronous CDMA Channels

~ince the Co.de Division Multiple Access (CDMA) system was proposed for the next generation of mobile communications, several multiuser detection· methods have been develope? to cope with the Multi-Access Interference inherent in the CDMA system. Fading and. multIpath ch~acteristi~sof the mobile radio channel present significant challenges to the designer of multiuser receivers. The fading phenomenon prevents the traditional Phase-Lock Loop (PLL) carrier recovery, so for the coherent detection, more accurate channel estimation is required. In this paper, we consider only frequency non-selective fading channels. Removing the usual assumption that the channel coefficients are perfectly known to the receiver, we model the fading channel as a second order Auto-Regressive (AR) process, and use the decision-directed Kalman filter to estimate the channel coefficients. The effect of estimation errors on the performance of several multiuser detectors is investigated by analysis and simulations. Issues concerning the implementation of the Kalman filter are also addressed.