A new blind signature waveform estimation approach in antenna array CDMA systems

We propose a new blind desired users' signature waveform estimation approach in antenna array CDMA systems. Unlike many existing blind channel identification methods based on subspace, relying on the orthogonality between the signal and noise subspace and a certain channel constraint, the proposed method directly obtains the total channel convolution matrix of all users by exploiting the cyclostationarity of received signals. It is more accurate at a little higher signal-to-noise ratio (SNR). We find out the solution to the desired users' signature waveform necessary for multiuser detection (MUD) given the estimated channel convolution matrix. The simulation results for MIMO systems show that the proposed method has superior performance over some existing methods at a little higher SNR.