WLS algorithm for a Complex Lattice Linear Predictor for Frequency Estimation and Tracking

This paper presents a simple lattice-structure complex adaptive linear predictor (LP) to estimate and track an unknown frequency of a noisy complex sinusoid. Instead of a widely used LMS-based algorithm, a weighted least squared (WLS) algorithm is proposed to improve the convergence rate of the predictor. Analytic results for asymptotic bias is derived in closed form and demonstrated that the WLS is asymptotic unbiased estimator.