The estimate of signals parameters is very important in wireless communications. In this paper, we combine subspace-based blind channel estimation algorithm with the extension of the JADE-WSF algorithm to jointly estimate the Angles-of-Arrival (AOAs) and delays of multipath signals arriving at an antenna array in Code Division Multiple Access (CDMA) systems. Our approach uses a collection of estimates of a consistent chip-sample of space-time vector channel. The channel estimates are assumed to have constant path AOA and delay over a finite number of symbols. Unlike the traditional MUltiple SIgnal Classification (MUSIC) and Estimation of Signal Parameters via Rotational Invariance Techniques (ESPRIT) algorithms for the estimation of signals parameters, the proposed method can work when the number of paths exceeds the number of antennas. The Cramer-Rao Bound (CRB) and simulations are provided.