Distributed Strain Sensing Based on Sensing Range Enhanced Optical Frequency Domain Reflectometry by Modified Longest Common Substring Algorithm

Conventional cross-correlation is commonly used in the estimation of the spectral shift for optical frequency-domain reflectometry (OFDR) based distributed sensing. However, the potential errors in cases of relatively large shifts have severely limited the sensing dynamic range. We report a modified longest common substring algorithm that exploits two-dimensional image processing for more accurate shift estimation. Theoretical study for the conventional cross-correlation is conducted, which reveals quantitatively the underlying mechanism for the emergence of the correlation errors. By adopting the relative error function as the similarity function in connection with the binarization and normalized projection, it allows for a reduction for the possibility of large estimation error in a two-dimensional image processing manner, leading to a substantial improvement in the sensing range. Experimental demonstrations have verified a remarkable enhancement of about 18.7 times in strain sensing range, which accounts for ∼56.2% of the effective sweep range. The proposed algorithm not only permits opportunities in OFDR based distributed sensing, but also promises enhanced fidelity for applications where cross-correlation is concerned.