Frequency estimation and tracking of multiple sinusoidal components

The paper presents a discrete-time algorithm for the frequency estimation of signals containing two independent sinusoidal components. The magnitudes and the phases of the components are obtained as well as their frequencies, and extension of the scheme to a higher number of components is possible. The algorithm is analyzed in the case of a single component using a linear time-invariant approximation. The results give estimates of the convergence rate and of the noise sensitivity of the algorithm. Estimation of two frequencies is performed by combining two estimators in a parallel structure. To prevent the two estimates from converging to the same frequency, a separation algorithm is proposed. Simulations show that the resulting estimator is effective at discriminating between two closely-spaced frequencies. Tracking of slowly-varying frequencies is also demonstrated.