Study on sliding-window length based on Rayleigh backscattering spectrum correlation in distributed optical-fiber strain measurement

A theoretical model is established for estimating the strain measurement error based on the Rayleigh backscattering spectrum correlation in distributed optical fiber strain measurements. Assuming the signal is much larger than the noise, the theoretical model predicts the strain measurement error using noise variance and the defined quality factor Q of the Rayleigh backscattering spectrum. Furthermore, an algorithm based on the quality factor Q is proposed to select an optimized sliding-window. The sliding-window length can be obtained by calculating the threshold value of the quality factor using a theoretical model corresponding to the required strain measurement accuracy. Compared with the traditional method where the sliding-window length is defined by the user based on spatial resolution requirements or an empirical definition, the sliding-window length determined by the algorithm is more reasonable and can be automatically defined, alleviating the requirement for user inputs. To verify the correctness of the theoretical model, two experiments are set up: a self-correlation experiment, that analyzes the effect of the quality factor on the strain measurement accuracy, and a virtual experiment of the noise influence, which analyzes the effect of different noise variances. The experimental results are in good agreement with those of the model.