An analytical blind multiuser detection for CDMA signals in multipath channels
Murat Torlak · 2002
The detection of CDMA signals in multipath channels can be easily done with the aid of training sequences. In the absence of training sequences, a number of blind adaptive and subspace-based multiuser receivers has been proposed for equalization of dispersive CDMA channels. We propose an analytical solution to direct equalization of CDMA signals. The method does not require training sequences or knowledge of interfering user signature waveforms and channel parameters. The method exploits the constant modulus property of CDMA signals and Toeplitz filtering matrix constructed from the desired user's spreading code. It requires less data samples than the blind adaptive multiuser receivers to converge. Thus, it is suitable for estimating initial receiver parameters before switching to decision directed mode. As a byproduct, the method estimates the channel parameters for the user of interest. It also provides an exact linear zero-forcing solution in the absence of noise.