Joint estimation of correlated multi-carrier DS-CDMA mobile fading channels based on optimal filters

Abstract

This paper deals with the joint estimation of rapidly time-varying and correlated Rayleigh fading channels in synchronous multi-carrier direct-sequence code-division-multiple-access (MC-DS-CDMA) systems. Usually, when the multiple carrier fading channels are modelled by autoregressive (AR) processes, they can be estimated separately by means of an optimal Kalman filter. However, a loss in performance can be expected when the channels are correlated. To take into account these correlations, the multiple carrier fading processes are stored in a vector, modelled as vector AR process, and estimated jointly by means of an optimal Kalman filter. Nevertheless, this requires the simultaneous estimation of the AR parameter matrices in the vector AR process. To avoid a non-linear approach such as the extended Kalman filter (EKF), this estimation issue can be solved by using dual Kalman filters. A comparative study on channel estimation is carried out between the proposed joint estimation scheme, the separate estimation counterpart and the standard least mean square (LMS) based estimator.