Enhanced vibration recovery of φ-OTDR under noisy circumstances using a signal-noise separation algorithm

Phase-sensitive optical time-domain reflectometry (Φ-OTDR) enables multipoint detection along optical fibers and has been widely applied in distributed acoustic sensing. However, due to the fundamental sensing mechanism, the target signal and ambient noise are inherently superimposed and simultaneously detected. This sensing characteristic results in a low signal-to-noise ratio (SNR) for reconstructed vibrations under noisy circumstances, particularly facing the strong ambient noise and spectral overlap. To address this issue, we propose an SNR enhancement algorithm based on the blind source separation theory. With this algorithm adopted, target vibrations can be quantitatively recovered in conventional Φ-OTDR schemes even under noisy circumstances with severe spectral overlap. The experimental results show that the SNR improvement from 9.4 dB to 36.6 dB for a 40-Hz vibration signal has been achieved. Furthermore, the algorithm successfully reconstructs both square-wave and triangular-wave vibrations from strong noise, achieving high-fidelity waveform recovery with correlation coefficients of 0.97 and 0.98 relative to the applied signals, respectively. These results verify the algorithm's robustness and effectiveness in reconstructing various vibrational patterns under severe noise aliasing.