The performance of time-lagged correlation and Fourier phase shift analysis in resonator-based optical sensors was evaluated and compared to centroiding algorithms. 100 emission spectra of an Er3+ oxyfluoride microsphere were recorded and the average of all data was used as a noiseless spectrum. The Whispering Gallery Mode (WGM) signal, superimposed on the Er3+ emission band, was subtracted and shifted to simulate sensor measurements. Additive random Gaussian noise with different values of the variance was added to the original and WGM-shifted spectra, and the WGM shift was estimated using time-lagged correlation, Fourier phase shift analysis and centroiding algorithms. The procedure was repeated 10000 times to evaluate the accuracy and precision of all numerical methods, and different values of the noise level and WGM shift were considered. These pseudo-experimental studies show that the use of time-lagged correlation and Fourier phase shift analysis can improve the accuracy and precision of the sensor when compared to traditional centroiding algorithms. Additionally, signal analysis can be automated without the need for peak identification and matching.
Read full abstract