Detecting target motion by frequency-plane smoothing

The technique of diagonal smoothing, which provides a consistent estimator of the two-dimensional spectrum for a harmonizable, periodically correlated (PC) random process, is extended to radial and Doppler smoothing to produce consistent estimators for the two-dimensional cross-spectrum of sensor outputs receiving a signal from a moving target. Simulations are presented to illustrate both the radial and the Doppler smoothing techniques. Since the diagonal smoothing readily adapts to produce cross-spectral estimators for sensor outputs governed by the linear motion model and consistent spectral estimation of almost PC processes, it is conjectured that radial and Doppler smoothing estimators will apply to these more general random processes.< >