A fast adaptive notch smoother for parameter estimation of complex sinusoid under noise

This paper presents an adaptive notch smoother for detection and extraction of a time varying complex sinusoidal signal buried in noise. Adaptive notch smoother is found to have better signal estimation accuracy than that obtained from adaptive notch filters (ANF). The ANS algorithm provides estimated frequency, amplitude and phase of the signal. But the basic problem of ANS is, for the estimation of smoothed amplitude parameter, the algorithm needs the value of smoothed frequency component. Hence the estimation of amplitude component is not simultaneous. To overcome this problem the proposed method combines adaptive Gauss-Newton algorithm with the ANS algorithm, for the simultaneous estimation of all the signal parameters.