Spectral correlation analysis of complex data

Abstract

Phase information of a coherent signal is recorded and preserved as complex data. Although Fourier spectrum has long been proven to be effective, it is sometimes difficult to properly estimate phase information from severely corrupted signals. We present and define two useful parameters: spectral correlation coefficient (SCC) and spectral covariance to mean power ratio (SCR). A simulation test was carried out using a random Gaussian phase noise. The SCC and SCR decrease linearly with extremely low gradients as the noise level increase, and especially the SCC maintains high values larger than 0.9 up to 80% noise corrupted signals. We applied the SCC method to a radar interferometric phase, and demonstrated that it was effective even for severely corrupted signals by noises.