A Krylov subspace approach to blind channel estimation for CDMA systems

The estimation of multipath channel parameters for CDMA systems can be easily done with the aid of training sequences. In the absence of training sequences, a number of subspace-based methods have been proposed for the estimation of CDMA channel parameters under frequency-selective fading. However, these methods generally rely on signal and noise space eigenvectors obtained by eigenvalue or singular value decompositions (EVD or SVD). In this paper, a blind channel estimation technique based on the Krylov subspace of the received CDMA signals without relying on EVD or SVD is proposed. Another advantage of this technique is that it can estimate the channel parameters using fewer subspace vectors as compared to the eigenvector based methods.