Blind adaptive multiuser detection in multipath CDMA channels based on subspace tracking

A blind adaptive technique for signal demodulation in multipath code-division multiple-access (CDMA) communication channels is proposed. This technique is based on signal subspace estimation. The receiver employs a bank of linear filters (decorrelating filters or linear MMSE filters) at the front end to mitigate the multiple-access interference and the multipath interference. A channel estimator is used to estimate the channel state for diversity combining. It is shown that through the use of signal subspace estimation, both the decorrelating filterbank and the linear MMSE filterbank can be obtained blindly, i.e., they can be estimated from the received signal with the prior knowledge of only the signature waveform of the desired user. Two forms of the subspace-based linear filterbanks are developed and their equivalence in terms of the interference suppression capability is established. These subspace-based blind adaptive interference suppression techniques require, at each symbol epoch, the eigenvalues and the eigenvectors of an appropriate signal subspace, which ran be obtained using computationally efficient sequential adaptive eigendecomposition (subspace tracking) algorithms. Moreover, a blind adaptive method for estimating the channel state is developed, which also produces the postcombining decision statistic as a byproduct.