Semi-blind CDMA code estimation in the downlink

Signal processing in the CDMA downlink is a challenging task since the codes of the disturbing users are unknown. Blind source separation methods could be used for this task, but they fail when the fading process is Gaussian, which is often the case in CDMA. In this paper, we try to solve the problem by using a semi-blind approach: only the code alphabet is assumed to be known, i.e. s_k∈{-1,1}, but the codes themselves are unknown. We introduce an iterative subspace algorithm which is used for code estimation. Simulations show that the convergence of the algorithm is fast compared to the more straightforward approaches. In addition, the algorithm yields very good estimates to the codes and their delays