BOTDR Denoising by Sparse Representation Algorithm with Preformed Dictionary

Abstract

In Brillouin optical time domain reflectometers, the signal-to-noise ratio is a key factor restricting the sensor performance. Using redundancy and correlation of 3DBrillouin gain spectrum in multi-dimensional domain, sparse representation algorithm can be used to improve signal-to-noise ratio. According to basic principle of sparse representation, a dictionary can be designed to reconstruct valid signals. During reconstruction, random noise will be discarded as residuals. In this paper, discrete cosine transform algorithm is used to design the dictionary, orthogonal matching pursuit algorithm is used to extract the coefficient matrix, and the signal is finally reconstructed to achieve the purpose of noise reduction. The simulation results show that when 5dBm random noise is added, signal-to-noise ratio in the non-temperature-change region is increased by 24.3dB, which provides a new idea for improving signal-to-noise ratio of BOTDR sensor.