Skip to main content

Chad Spooner: Introduction to Cyclostationary Signal Processing for Blind Signal Detection and Characterization

The presence of man-made RF interference (RFI) in radio-astronomy observation bands is an ongoing problem and is likely to worsen with the increasing use of the RF spectrum and satellite-communications technology around the globe and in space. Determination of the types and parameters of interferers can assist in determining their origin, which can help in efforts to force transmitter owners to vacate reserved bands. Such characterizations can also provide valuable information to interference-mitigation (signal-separation) techniques that attempt to remove the offending signal while leaving the naturally occurring signal of interest undisturbed. The detection,
characterization, and identification of RFI can be considered as a variant of a more general problem called RF scene analysis (RFSA), in which all sources of man-made RF energy in any band are to be blindly detected and characterized. RFSA problems arise in spectrum monitoring, surveillance, cognitive radio,and military settings. In this talk, we present a general framework for RFSA that exploits the fundamental statistical structure of RF communication signals: cyclostationarity. Cyclostationary signals stand in contrast to stationary signals in that their probabilistic parameters, suchas mean and autocorrelation, are not time-invariant but are periodic functions of time. A tutorial overview of the theory of cyclostationary signals is sketched, a comprehensive RFSA algorithm is described, and results obtained by applying cyclostationary signal processing to GBO-supplied RF data sets are presented.