Doppler radar signatures analysis by using joint bispectrum-based time-frequency distributions

Estimation of instantaneous frequency (IF) time-varying behavior of a non-stationary and multi-component signals embedded in additive Gaussian noise is considered for Wigner-Ville (WVD), Wigner bispectrum (WBD), parametric (PBBD) and non-parametric (NPBBD) bispectrum-based distributions. A performance comparative study between WVD, WBD, PBBD and NPBBD is carried out by computer simulations both for several non-stationary and multi-component test signals and real radar backscattered echo-signals measured by Doppler surveillance radar for a moving human. Analysis of time-frequency (TF) distributions shows that bispectrum-based approach permits to detect and extract phase coupling among pairs of Doppler IFs containing in non-stationary and multi-component wideband signals. Experimental radar micro-Doppler signatures derived from the returns measured for moving human demonstrate important information features about the object dynamics and kinematics. These information features can be useful in radar automatic target recognition systems.