RFI subspace smearing and projection for array radio telescopes

Abstract

Active Radio Frequency Interference (RFI) mitigation becomes a necessity for radio astronomy. The solution commonly applied by the community consists in monitoring the statistics of the received signal, and flag out the detected corrupted data. Subspace projection with array radio telescopes has been suggested as an alternative to data excision to avoid important losses of data and overcome its inherent ineffectiveness with continuous interference. Spatial filtering relies on the estimation of the RFI spatial contribution, and the projection of the subspace spanned by the RFI out of the observed data vector space. To perform well, the dimensionality of the RFI subspace is constrained. RFI subspace estimation techniques assume the source of RFI to be spatially stationary over the sample covariance matrix evaluation. When the relative movement between the telescope and the interferer becomes significant, the RFI subspace gets smeared over the whole data vector space. The subspace projection can then no longer be applied without affecting the source of interest recovery. This paper addresses the effect of RFI subspace smearing on the subspace projection approach, and suggests an alternative technique based on a covariance matrix subtraction, improving the performance of spatial filtering in the case of high subspace smearing.