Performance enhancement of sub-sampling Φ-OTDR sensing through kurtosis statistics algorithm

Abstract The frequency response and spatial resolution (SR) of the phase-sensitive optical time domain reflectometry vibration sensing are constrained by the sensing distance and optical pulse width, respectively. Non-uniform sub-sampling has been employed to facilitate the measurement of high vibration frequencies across extended sensing distances breaking through the sampling theorem. However, this often results in a reduced signal-to-noise ratio (SNR) due to the low sampling rate. To address these limitations, we propose a novel vibration signal recognition algorithm combining the moving average and moving differential (MA-MD) algorithm with the kurtosis statistics scheme, and it is experimentally validated by recognizing vibration signal detected by the sub-Nyquist additive random sampling. A 100 kHz vibration signal at about 4.8 km was successfully detected with an average sampling rate of 4.04 kHz and a 400 ns pulse width, yielding an SNR of 9.82 dB and an SR of 6.8 m. It is also demonstrated that this signal recognition approach substantially outperforms the individual MA-MD and kurtosis statistics methods.