A cumulant-based method for the direct estimation of the spatial Wiener filter

In this paper we propose a new algorithm for performing blind source separation and signal recovery for the narrowband case. This algorithm seeks to estimate and implement the Wiener filter directly, using the second- and fourth-order cumulants. The estimated filter is asymptotically consistent with or without noise, and is not subject to the bound imposed by pre-whitening as with some existing algorithms. In many cases it nearly achieves the clairvoyant Wiener limit as performance studies show.