Spatial filtering of RF interference in radio astronomy

We investigate spatial filtering techniques for interference removal in multichannel radio astronomical observations. The techniques are based on the estimation of the spatial signature vector of the interferer from short-term spatial covariance matrices followed by a subspace projection to remove that dimension from the covariance matrix, and by further averaging. The projections will also modify the astronomical data, and hence a correction has to be applied to the long-term average to compensate for this. As shown by experimental results, the proposed technique leads to significantly improved estimates of the interference-free covariance matrix.