Adaptive maximum windowed likelihood AM-FM signal decomposition

Abstract

A maximum windowed likelihood (MWL) criterion is suggested to adaptively estimate the amplitudes and the frequencies of the components of a real signal composed of multiple sinusoids. We extract the amplitudes using the MWL criterion, then a gradient-based adaptive method is employed to track the frequencies. The proposed algorithm is implemented using the parallel modules with low computational complexity. Simulations have shown that the algorithm has a high frequency resolution. The effect of the window (length and type) on the behavior of the algorithm is investigated. The relationship between the lock-in range and the window type illustrates that the algorithm can be efficiently used in the different environments.