Modeling of multicomponent AM-FM signals using FB expansion and linear TVAR process

In this paper, we propose a novel method of modeling multicomponent amplitude and frequency modulated (AM-FM) signals using linear time varying autoregressive (TVAR) process. In the proposed method, the Fourier-Bessel (FB) expansion decomposes the multicomponent AM-FM signal into a number of single component signals and a 2-order TVAR process is used to model each single component signal. The estimation of the TVAR parameters of a multicomponent signal requires the inversion of a large correlation matrix, whereas the proposed method requires the inversion of a number of relatively small correlation matrices with better numerical stability properties. In presence of noise, it is demonstrated that our method improves the estimation of the time varying poles. Simulation results are presented for two-component damped AM sinusoidal signal and two-component damped FM sinusoidal signal.