RFI detection by automated feature extraction and statistical analysis
Benjamin Winkel; Argelander Institute for Astronomy (AIfA) University of Bonn Auf dem Hügel 71, D-53121 Bonn, Germany
Abstract:
Modern FPGA based spectrometers provide not only superior input-signal quantization (of at least 8 Bits) but allow to store spectra on hard disks within less than a second. This high temporal resolution of the data gives us for the first time the opportunity to apply sophisticated off-line radio frequency interference (RFI) detection schemes to the data. We present an automated feature identification and extraction algorithm based on two-dimensional baseline fits, followed by statistical and morphological analyses of all spectral features. Simulations confirm, that our detection procedure allows to identify the absolute majority of RFI signals down to the needed 3-sigma signal-level.
The RFI detection software was applied to real observational data of the 100-m telescope at Effelsberg. We show, that a large fraction of the detected RFI events are produced from accurately operating devices, localized within the observatory building itself. This emphasizes the necessity of proper shielding of all electronic devices in the immediate vicinity of a radio telescope.