Improvement of FBG peak wavelength demodulation using digital signal processing algorithms

The main feature of Fiber Bragg Gratings is that they perform a direct transformation of measured parameter into a shift in the center Bragg wavelength. Wavelength shift of reflected light is independent of optical source power changes. Spectra reflected or transmitted by FBG illuminated by high power source of light usually has smooth shape similar to Gaussian curve, with high signal to noise ratio. In engineering application, fiber Bragg grating structure contains e.g. electronic elements and is placed in the environment which make the measured signal contain various types of noise which seriously influences the precision of demodulation, especially where sensing array is made by large number of sensors or while is used broadband, low power source.
This paper presents a possibility for extraction FBG peak wavelength from spectra with strong noise with most frequently using digital signal processing algorithms. Spectra of tested FBG were measured in 28 points with different extending force applied to tested sample. The shape of sample provides an equal strain in full longitude of FBG. It allows to compare the processing characteristics received in two different ways: from reference spectra measured with using SLED source and spectra traced with using halogen source which containes high influence of noise. The comparison of function minimum, centroid and gaussian fitting methods for peak wavelength detection is included. Digital filter and various methods of center Bragg wavelength detection are jointly applied to improve wavelength detection accuracy. The linearity of processing characteristics of extended FBG measured for reference high power and second, low power source is compared.