PERFORMANCE ANALYSIS OF A CORRELATION-BASED SINGLE TONE FREQUENCY ESTIMATOR

This paper analyzes the frequency error variance of a low complexity s ingle tone frequency estimator based on sample correlations of the inp ut data. In the high SNR scenario it is analytically shown that the ac curacy of a properly tuned algorithm is nearly optimal, i.e. nearly at tains the Cramer-Rao lower bound. For low SNR the statistical efficien cy of the algorithm is degraded, but it is analytically proven that fo r a large number of samples the error variance attains the lower bound for this class of estimators.