USE OF FREQUENCY TRACKERS IN LASER DOPPLER VELOCIMETRY FOR SOUND FIELD MEASUREMENT: COMPARATIVE STUDY OF TWO ESTIMATORS

Abstract This paper deals with the processing of signals resulting from the measurement of sound fields using the laser Doppler velocimetry technique. For such an application, the problem lies in the demodulation of signals whose instantaneous frequencies have small and fast variations. Two frequency trackers are systematically compared using Monte-Carlo simulations. The first tracker is parametric and is an adaptation of Kalman filtering, while the second is non-parametric and based on the cross Wigner–Ville distribution. As opposed to time–frequency techniques, Kalman filtering does not suffer from a limiting short-time assumption; its adaptability makes it able to accurately track any instantaneous frequency variation in a noiseless environment, even when very low modulation indices are involved. For low signal-to-noise ratios, the Wigner–Ville-based estimator performs better; for moderate signal-to-noise ratios (higher than 12 dB), Kalman filtering represents the best compromise in terms of bias and variance.