A subspace-based blind interference cancellation is proposed. It is shown that the performance of the proposed detector is the same as that of the decorrelating detector. The aymptotic results of the proposed detector and its near-far resistance are examined. The proposed subspace-based blind interference cancellation outperforms the previous blind minimum-output energy (MOE) detector. We also propose an improved natural power method. Then a blind adaptive implementation based on this method is developed. The improved natural power method is the signal subspace tracking algorithm mainly adequate for the subspace-based blind interference cancellation which does not need estimation of eigenvalues. The proposed blind adaptive detector can detect desired user`s bit signal with the prior knowledge of only the signature waveform, timing and amplitude of user of interest. It also outperforms the previous blind adaptive RLS MOE detector.
