Phase-sensitive optical time domain reflectometer with ultrafast data processing based on GPU parallel computation

The sensing performance of a phase-sensitive optical time domain reflectometer (ϕ-OTDR) has been sufficiently improved, thanks to plenty of valuable research in recent years. However, in the literature, there is hardly any attention aimed at enhancing the data processing capability of the system, the necessity and significance of which are undisputed. This paper, for the first time to the best of our knowledge, analyzed the intrinsic superiority of employing GPU parallel computation in ϕ-OTDR for improving the data processing capability and presented a comprehensive performance evaluation. Three typical, frequently implemented algorithms in ϕ-OTDR-moving average, batch fast Fourier transform, and batch correlation dimension computation-are carried out where CPU-based programs and counterpart GPU-based programs are, respectively, developed. Their time-consuming performances in different data scales are experimentally tested and compared. The experiment results show that in each case, employing GPU parallel computation can significantly enhance the system's data processing capacity, thus providing a feasible and efficient way of guaranteeing real-time operation with the growing data scale.