Research on fast and accurate extraction method of Brillouin frequency shift based on window-weighted centroid algorithm

Curve fitting method (CFM) is the primary method for extracting Brillouin frequency shift (BFS) in Brillouin sensing system. However, the accuracy of extraction result is affected by the initial parameter settings, the spectral shape of the Brillouin power spectrum (BPS) and the fitted curve model. Additionally, the fitting process is time-consuming. In view of this problem, this paper proposes a fast and accurate extraction method of BFS based on the window-weighted centroid algorithm (WCA). This method achieves extraction of BFS by selecting data window and calculating weighted centroid for BPS. The proposed method was experimentally verified using a Brillouin Optical Time Domain Reflectometer (BOTDR) temperature sensing system for extracting the temperature information of the 50 m heating section near the end of a 9.4 km optical fiber. Compared with CFM, cross-correlation method (XCM), backward propagation neural network (BP-NN) and centroid algorithm (CA), the proposed method has a higher accuracy and faster processing speed and is suitable for dealing with distorted non-ideal spectral shape. The proposed WCA method is an alternative method for extracting BFS information, especially in situations where the real-time requirement is high and the spectral shape is complex.