Discriminating Interference Fading Locations in Φ-OTDR Using Improved Density Clustering Algorithm.

Area of Science:

• Optical Engineering
• Signal Processing
• Data Science

Background:

• Phase-sensitive optical time-domain reflectometer (Φ-OTDR) systems enable real-time, distributed sensing of weak vibration signals.
• High laser coherence in Φ-OTDR enhances sensitivity but introduces fading noise, causing phase demodulation distortion and compromising system reliability.
• Interference fading is a fundamental challenge limiting the performance of Φ-OTDR systems.

Purpose of the Study:

• To propose and validate an optimized density clustering algorithm, adaptive principal component analysis DBSCAN++ (AP-DBSCAN), for mitigating fading noise in Φ-OTDR systems.
• To accurately identify and reconstruct data points affected by fading noise, thereby improving the reliability of vibration measurements.
• To enhance the computational efficiency of fading noise mitigation in Φ-OTDR.

Main Methods:

• Identification of fading regions based on the fading principle.
• Adaptive determination of DBSCAN parameters (eps and Minpts) using K-distance integration.
• Application of Principal Component Analysis (PCA) and DBSCAN++ for efficient and accurate fading point detection.
• Reconstruction of compromised data points using nearest-neighbor interpolation.

Main Results:

• The proposed AP-DBSCAN method achieved a high fading-point detection accuracy of 99.92%.
• Computational efficiency was significantly improved, ranging from 67.33% to 76.29% compared to existing methods.
• Experimental results demonstrated superior performance over standard DBSCAN, FDBSCAN, and DBSCAN++ algorithms.

Conclusions:

• AP-DBSCAN effectively addresses the challenge of interference fading in Φ-OTDR systems.
• The algorithm provides adaptive parameter determination and accurate data reconstruction, enhancing system reliability.
• The method offers a significant improvement in both accuracy and computational efficiency for Φ-OTDR applications.