Low complexity block-based unbiased frequency estimation

The problem of efficient block-based frequency estimation is addressed in this paper. Different to a recently proposed solution that uses a direct-form realization of the notch filter, the novel estimator is based on the normalized lattice realization. One specific advantage of the proposed realization is that no additional normalization is required, as a consequence, a low complexity estimator is obtained. The mean squared convergence is probed and an analytic expression for the Mean Squared Error for infinite data is given. Simulations show that the algorithm converges in a few iterations. Comparisons with a previous block based estimator are also shown.