STNet: A Time-Frequency Analysis-Based Intrusion Detection Network for Distributed Optical Fiber Acoustic Sensing...

Area of Science:

• Engineering
• Signal Processing
• Cybersecurity

Background:

• Distributed optical fiber acoustic sensing (DAS) is crucial for long-distance intrusion detection.
• High-intensity interference noise in realistic environments significantly degrades DAS performance.
• Existing methods struggle to maintain detection accuracy under noisy conditions.

Purpose of the Study:

• To propose and evaluate STNet, a novel intrusion detection network for DAS systems.
• To leverage the noise-resistant properties of the Stockwell transform (S-transform) for improved disturbance detection.
• To enhance intrusion detection rates and reduce false alarm rates in complex, noisy environments.

Main Methods:

• Utilizing the Stockwell transform (S-transform) to extract time-frequency features from DAS signals.
• Processing space-time data matrices derived from DAS signals using a sliding window approach.
• Employing a non-maximum suppression algorithm (NMS) for precise intrusion localization and post-processing.

Main Results:

• STNet demonstrated satisfactory performance in detecting intrusions within a high-intensity noise environment.
• The proposed method effectively mitigated the impact of interference noise on DAS detection.
• Experimental validation in a realistic high-speed railway setting confirmed the method's efficacy.

Conclusions:

• STNet offers a robust solution for intrusion anomaly detection in DAS systems, even under challenging noise conditions.
• The integration of S-transform and STNet significantly enhances detection accuracy and reliability.
• This approach provides an effective strategy for achieving high intrusion detection rates and low false alarm rates in complex scenarios.