Abstract: We propose a method of adaptive signal decomposition based on the linear frequency modulated Gaussian elementary functions and the related adaptive time-frequency distribution that is positive, cross-term interference free and with better time and frequency resolution. The set of the elementary functions is obtained by the successive matching of the signal local nature with the single prototype Gaussian function through regulating its variance, sweep rate, translation and modulation in an adaptive manner. The matched coefficients and the parameters contain all the information of the signal and can be used for the further processing of the signal. The effectiveness of the method has been proved by simulation results.

Key words: linear frequency modulated Gaussian function, nonstationary signal, adaptive decomposition, time-frequency distribution