A robust spectral estimator based on the smoothed Wigner-Ville distribution

Summary form only given. The Wigner-Ville distribution (WVD) presents excellent time and frequency resolution, which make it very attractive for signal analysis. Smoothing the WVD (SWVD) in time reduces the interference between frequency components. The authors have proved that for noise-free real or analytic signals, the frequency estimates obtained from the poles of the autoregressive (AR) polynomial are exact and independent of both the phase and data length. Furthermore, they have shown that the SWVD half-kernel used in this work also produces precise AR frequency estimates, regardless of phase and window length, and has the additional advantages of high temporal resolution and excellent performance in the presence of noise. The theoretical results obtained for monocomponent signals have been extrapolated to multicomponent signals in noise