Improved weighted phase averager for frequency estimation of single sinusoid in noise

Proposed is an improved weighted-phase-averager frequency estimator which is obtained following the approach of Kay (see Reference [1]), but incorporates the present authors' new phase noise observation model developed (see Reference [2]). Their mean square estimation error performances are compared using computer simulations, the results of which show that the estimator based on the new model performs moderately better, especially at low signal-to-noise ratios. Moreover, since the phase noise in the new model has a time-varying variance, it can intelligently identify those observed instantaneous signal phase samples with larger errors. By dropping these samples, it is shown by simulation that the improved weighted phase averager can lead to further performance improvement for low SNR.