Recognition of radar signals by time - frequency peak filtering

In this paper the recognition of radar signals by the time - frequency peak filtering (TFPF) is considered. The term recognition in this case can be understood as the reconstruction of signal waveform from observations corrupted by additive noise. Time - frequency peak filtering method (TFPF) is based on encoding of the noisy signal as the instantaneous frequency (IF) of a unit amplitude frequency modulated (FM) analytic signal. Estimation of the IF encoded signal, performed by taking the peak of the pseudo Wigner - Ville distribution (WVD), allows to get the estimation of signal waveform. It is shown that the length of the window of the pseudo Wigner - Ville distribution has a great significance for the accuracy of the TFTP algorithm. The performance of the pseudo WVD TFPF is assessed on the real radar signal. It is showed Nearly clear reconstruction of the signal in the low SNR ratio is presented in this paper.