An adaptive notch filter for frequency and power estimation with application in telephone signaling codes

Adaptive second-order notch filters are presented for detection and tracking of multifrequency signals corrupted by noise. The adaptation process of the notch was made from the complex coefficients of the first-order linear prediction error filter, associated with the notch. The theoretical analysis yields simple results which are useful for the detection of sinusoidal components in a spectrum. The least-squares algorithm presented does not require matrix inversion, making it easy to compute. The method makes it possible to obtain the power of the input sinusoid and an accurate frequency estimation. The authors apply the method to MFC and DTMF (dual-tone multifrequency code) digital telephone signaling, and obtain useful results for real-time practical applications.<>