Blind adaptive multiuser detection in multipath code division multiple access channels

In this paper, blind adaptive multiuser detection in multipath CDMA channels based on the MMSE (minimum mean-squared error) criteria is proposed. Compared with the multiuser detections presented by M. Honig et al. (see IEEE Trans. on Information Theory, vol.41, no.4, p.944-60, 1995) and X.D. Wang and H.V. Poor (see IEEE Trans. on Information Theory, vol.44, no.2, p.677-90, 1998; IEEE Trans. on Signal Processing, vol.46, no.11, p.3030-44, 1998), this one takes advantage of the structure and characteristic of two subspaces: 1) the subspace structure associated with the code of the desired user; 2) the estimated "signal subspace" of the covariance matrix for the observed signals. Simulations state this detection has strong ability against near-far and multipath effects. Meanwhile, it avoids significant performance reduction when signature mismatch exists.