On the recovery of moving source characteristics using time–frequency approach

Four time–frequency analysis techniques, namely the short-time Fourier transform, the wavelet transform, the polynomial phase estimation and the Chirplet transform, are used in the present study to recover the moving speed, the sound frequency and the strength of a point source moving with a subsonic speed relevant to those of the road vehicles. Their performances in the presence of background random noises are tested in detail. The instant of the highest rate of change of the Doppler frequency is used as the time reference in the parameter recovery process. Results suggest that the performances of short-time Fourier transform and the wavelet transform, especially the former, are not satisfactory when compared to those of the other two methods even when the signal-to-noise ratio is reasonably high. The Chirplet transform gives a performance which is the least affected by the signal-to-noise ratio, while the accuracy of the polynomial phase estimation is the best among the four methods tested.