Optimal characterization of time-frequency measurements on distorted nonstationary signals

The problem of estimating the instantaneous frequency and amplitude for nonstationary noisy signal is addressed. An analysis of the casual errors due to the presence of noise is given for Wigner-Ville distribution; particularly, the theoretical bias for instantaneous frequency and amplitude is carried out. The Authors have already developed suitable interpolation techniques for both STFT and SPWD distributions, with the aim of reducing the incorrect spectral estimates caused by frequency variation and lack in synchronization. So high levels of accuracy, flexibility and efficiency, have been obtained even if heavily nonstationary components were present. Now, the research carried on with the analysis of the stochastical disturbance, by examining closely their effects on the methods proposed, having the object of developing a modified version that produce high levels of accuracy also when only noisy input data are available. The final aim is to implement the suggest techniques in a dedicated digital instrument, to perform accurate time-frequency analysis of nonstationary distorted noisy signals.