A Lightweight Depthwise Separable Convolution Network for Event Detection in Phase-Sensitive Optical Time Domain Reflectometry

Key Points

• To develop a lightweight convolution network for effective event detection in optical time domain reflectometry.
• Utilized depthwise separable convolution to minimize computational load
• Applied the network to phase-sensitive optical time domain reflectometry data
• Performed accuracy assessments comparing traditional methods and the proposed network
• Achieved significant improvements in detection accuracy
• Reduced computational complexity compared to standard convolutional networks
• Demonstrated robust performance across different event types