Real-Time Anomaly Detection for Submarine Cables Based on Multivariate CUSUM
Chunying Xu, Jianrong Chen, Jingqi Fang, Keyan Xiao, Yuhong Xu, Jiawang Chen, Chuliang Wei · 2025
Security monitoring of submarine cables is essential for ensuring the reliability of transoceanic communications and power transmission. This paper presents a real-time monitoring method for submarine cables using an improved multivariate cumulative sum (MCUSUM) control chart. The method combines the sliding window technique with CUSUM control chart technology to detect abnormal fluctuations in submarine cable sensor networks. Experimental results show that the proposed method achieves a detection accuracy of 99.54% for 8 types of anchor smash events and 100% for 4 types of anchor hook events. For all 12 event categories, the false alarm rate remains below 2%, with an average detection latency of less than 0.2 seconds. This approach offers an effective solution for submarine cable security monitoring.