CFAR detectors for unknown signals based on the spectral correlation measurement

This paper deals with the unknown signal detection problem by using spectral correlation-based detectors. Spectral correlation-based techniques exploit the spectral redundancy present in manmade signals. However, most of them are limited to the analysis of the spectral correlation function in some specific regions of the frequency-cycle frequency plane, since some knowledge about the signal is assumed. In this work, two CFAR detectors are extended to the whole frequency-cycle frequency plane for the detection of completely unknown signals. Mathematical formulation and some radar signal examples are provided. Noise is assumed to be stationary, white, and zero-mean.