New multiple-model channel estimators for equalization of DS-CDMA fading mobile channels

For direct-sequence code-division multiple-access (DS-CDMA) time-varying channels with additive Gaussian noise, we propose a new receiver. Since channel tracking plays an essential role in practice in equalization of the receiver, a new and viable design approach, a modified interacting multiple-model (MIMM) algorithm, is proposed for tracking of multipath time-varying fading channels. We employ a set of second-order auto-regressive (AR2) multiple models associated with a Makov jump process to cover the range of channel dynamics. This setup leads to an AR2-based MIMM channel tracker which can automatically adjust channel model parameters to track time-varying fading channels for all times once some design parameters are initialized. We employ a Kalman filter as a symbol detector (KBSD). Simulation studies demonstrate that the performance of the AR2-MIMM channel tracker together with a conventional KBSD performs at least as well as that of the ideal AR2-based Kalman channel estimator having the exact knowledge of the Doppler rate of each channel in combination with the KBSD