Phase estimation using sine fitting with low number of samples is biased in the presence of additive noise

Abstract

It is demonstrated here that the initial phase estimation of a sine wave corrupted by additive noise using the least squares sine fitting algorithm is not biased, even for low number of samples, contrary to what happens in the case of the amplitude estimation. Monte Carlo simulations are presented which corroborate the conclusion reached. This result is of particular importance when studying the uncertainty of measurement methods based on phase estimation like ultrasonic non‑destructive testing.