EEG signal analysis using FB expansion and second-order linear TVAR process

In this paper, we propose a second-order linear time-varying autoregressive (TVAR) process for parametric representation of the electroencephalogram (EEG) signals. The coefficients of the Fourier–Bessel (FB) series expansion have been used to constitute a feature vector for segmentation of the EEG signal. Our approach is novel in the sense that by selecting an appropriate data length, we find a simple model for parametric representation of the EEG signals. The complete method for estimation of model parameters is presented in this work.