Sequential Frequency Estimation Using Auxiliary Particle Filter

Frequency estimation is an important and challenging task in signal processing and related applications. However, the measurement signals are contaminated with undesired noise sources in practice, and the true frequencies, hence, cannot be extracted directly. Particle filtering (PF) is a sequential Monte Carlo (SMC) method that is proved efficient and effective in estimating the frequencies from the timefrequency representations (TFRs). There are many techniques proposed to enhance estimation performance of PF. This work conducts a study on the performance in estimating frequencies from the spectrogram of the signal (computed using shorttime Fourier transform (STFT) using sequential importance resampling (SIR) PF and auxiliary PF (APF). APF is proved to providing superior performance in frequency estimation at low signal to noise ratios (SNRs).