In last decade, the distributed fiber sensors are attracting wide attentions for practical applications, such as Brillouin optical time domain reflectometer (BOTDR), which has been widely used in health monitoring of civil structure, safeguarding of pipeline and railway. However, the accuracy of Brillouin Frequency Shift (BFS) extraction suffers from the weakness of Brillouin scattering, and the traditional BFS extraction method is time consuming. These pose great limitations for BOTDR applications.
Recently, researchers at Shanghai Institute of Optics and Fine Mechanics, Chinese Academy of Sciences, theoretically and experimentally demonstrate two iterative method, iterative centroid analysis (ICA) and iterative quadratic fitting (IQF), to extract BFS from noise signals of BOTDR with fast processing speed.
Their analyses show that the accuracy of extracted BFS is deduced as a function of signal-to-noise ratio (SNR), frequency step, Brillouin linewidth, and the data window used in the BFS extraction.
Theoretically, it is found that the BFS variances are largely affected by the data window deviation (DWD). Thus, an iterative method was proposed to reduce the DWD caused error. In the first step, they took the frequency at maximum of spectrum obtained by averaging of several traces as the data window center xm. The trace number M in averaging could be adjusted according to the signal noise level. Following this, they carried out centroid analysis (quadratic fitting) for the M averaged spectrum with xm as data window center. A new peak frequency xc was then obtained. To the end, xm was replaced by xc and the second step was repeated until the extracted peak converges to a stationary value. The whole process can be displayed in the Figure 1.

Fig. 1 (a) The extracted centroids, (b) The extracted BFS by IQF and (c) its standard deviation.