Choice of Wavelet for Filtering of Signal from Fiber-Optic Temperature Sensor

One of the problems that arise during the design of a fiber-optic distributed temperature sensor is signal filtering. A high noise level occurs due to the optoelectronic conversion and the low power of the initial signal. A low signal-to-noise ratio leads to degradation characteristics of the fiber-optic temperature sensor. Averaging of multiple signals is often used to solve this problem. It significantly reduces device performance. Therefore, it is necessary to investigate and develop effective methods of signal filtering. In this paper, a filtering method based on the wavelet transform is considered. The method consists in splitting the signal into frequency bands by discrete wavelet transform and restoring the signal from the selected frequency bands. But at the same time, the choice of a wavelet plays a significant role. This paper discusses Bi-orthogonal wavelets, Daubechies, Coiflets, and Symlets. As a result, optimal wavelets were found for filtering the signal from the fiber-optic temperature sensor. An important advantage of this method is a slight decrease in spatial resolution with a significant reduction of noise level.