A canonical interferencelet-based approach to RFI identification

Abstract

Radio frequency interference (RFI) is a significant concern for radio astronomy, even in radio-quiet reserves. Whereas frequency-based algorithms are well suited for detecting and classifying intentional RFI (like communication links), the identification of unintentional RFI events (like those from devices such as relays or electric motors) is far more difficult. In this work we describe a novel way of dealing with time-domain unintentional RFI events. We show that irrespective of their sources, the time-domain RFI signals seem to consist of sub-impulses. These sub-impulses are akin to a canonical set of signals which we have named Canonical Interferencelets. We establish this hypothesis through a range of experiments. This method of analysing RFI signals is deemed very useful in the detection and identification of RFI events.