# (c) Time-frequency analysis algorithm

The signal character differences between a moving target and a stationary target can be revealed in the time-frequency domain. Motions of a moving target changes the Doppler centroid and Doppler modulation rate of the moving target, and time-frequency analysis algorithms are highly efficient in extracting the motion parameters. WVD transform is proposed by S. Barbarossa to extract and estimate the motion parameters of moving targets. After WVD transform, the azimuth spectrum is transformed into a straight line, the slope of the line is the Doppler modulation rate, and the frequency initial is the Doppler centroid. Suppose the signal is s(t), the definition of WVD transform is

The WVD transform has a poor performance in the low SNR condition, so the clutter suppression and spectrum extraction must be performed before WVD. Thus, WVD transform is more suited as a GMTIm algorithm but not a GMTI algorithm.

Furthermore, WVD transform is a bilinear transform, and is not suitable for multi-target detection and parameter estimation. Improved WVD transforms have been studied, while the performances are not promising as well. Wavelet transform is also a sort of new time-frequency analysis algorithms, but it has a heavy calculation burden and not suited for practical applications.