Parametric estimation of multicomponent signals using minimum cross entropy time-frequency distributions

The analysis and representation of multicomponent signals, embedded in additive Gaussian noise, is of interest in many signal processing applications and has been studied for years, mainly for the case of stationary signals. However, in the non-stationary case only a few methods are available, one of which is parameter analysis from the time-frequency distribution (TFD) of the signal. A new positive distribution free of cross-terms, named minimum cross entropy-TFD (MCE-TFD) was introduced previously. Based on this TFD, an algorithm for signal reconstruction from the MCE representation is introduced. The algorithm synthesizes the instantaneous frequency (IF), the phase φ(t) and the amplitude a(t) of each signal component and thus the entire original signal. Together with the MCE-TFD iterative approach, the proposed reconstruction algorithm provides a powerful tool for the representation and analysis of non-stationary multicomponent signals