CCA Colloquium: Gelu Nita
Speaker: Gelu M. Nita, Center for Solar Terrestrial Research, New Jersey Institute of Technology, Newark, NJ, USA
Title: The Generalized Spectral Kurtosis Estimator: A Review of Its Statistical Properties and Applications
Abstract: Owing to its conceptual simplicity and its proven effectiveness in real-time detection and removal of radio frequency interference (RFI) from radio astronomy data, the Generalized Spectral Kurtosis (SK) Estimator is likely to become a standard tool embedded in the hardware correlators or in the data processing pipelines of a new generation of radio telescopes. What makes a hardware or software SK correlator different from a traditional one is the fact that, in addition to the standard autocorrelation spectrum, which is accumulated to reduce the statistical fluctuations of the raw spectra, it also computes the SK estimator spectrum that has the remarkable property of having an unbiased unity expectation at all frequencies, if the input astrophysical signal of interest obeys Gaussian time domain statistics. Unlike natural signals, most of the artificial signals producing RFI obey different statistics. Thus, the SK estimator spectrum may be efficiently used to automatically flag the frequency channels whose estimators depart from unity, one way or another, by more than a set of predefined detection thresholds characterized by known probabilities of false alarm, which may be analytically computed according to the theoretical probability distribution of the SK estimator. Nevertheless, since it was first proposed to the radio astronomy community as an efficient RFI real-time detector, it has been shown that the practical applications of the SK estimator extend far beyond its originally intended scope. In this talk, I will present a review of the theory of the Generalized Spectral Kurtosis Estimator and discuss its possible applications as a statistical discriminator of natural and artificial signals, which will be illustrated by the results of an analysis involving two statistically different radio signals: a Fast Radio Burst pulse captured by EVLBI, and a deep-space artificial signal captured by the GBRT Breakthrough Listen digital recorder backend.
Please email [email protected] for Zoom information if you would like to participate.