An efficient technique for high correlation estimation

Two sum-based estimators are proposed as fast, accurate methods for estimating the normalized acf of stationary processes. For Gaussian processes, their performances are evaluated in terms of bias and covariance, and are compared to the ones of the conventional Direct estimator and of the Hybrid-Sign and the Polarity Coincidence fast methods. For high correlation values, the proposed estimators exhibit the same accuracy as the Direct estimator.