A time-varying optimal finite impulse response filter based estimation of short-time Fourier coefficients

초록

In this paper, a new short-time Fourier analysis method for quasi-periodic signals is proposed based on a time-varying optimal finite impulse response (FIR) filter. A new state space model of quasi-periodic signals is proposed by defining the stochastic model for the increment of a Fourier coefficient (FC) in order to deal with the biased variants of FCs. Then, the time-varying optimal FIR filter is applied to the proposed model to estimate the FC and its increments. Since the proposed state-space model and time-varying optimal FIR filter can be reduced to the previous state-space model and time invariant FIR filter, the proposed method can provide a more general solution than the existing FIR filter-based short-time Fourier analysis method. Through a numerical example, it is shown that the proposed method has better estimation performance and tracking ability for estimating FCs and their increments compared with existing approaches. © ICROS 2018.

키워드