Spectrally smoothed polynomial fourier transform for time-frequency characterization at low SNR

Frequency Modulated (FM) signals are encountered in many natural and human-made applications in earth, oceanic and atmospheric environments. In this paper, the limitations and optimality of existing FM signal characterization methods, in particular at low signal to noise ratios are investigated. For proper characterization, the selection of a time-frequency representation through spectrally smoothed Polynomial Fourier transform is proposed. The efficiency and the optimality of the method are elaborated.