Research on submarine cable detection technology based on information fusion

Xuejun Zhou, Shuicheng Yan, Ming-jun Zuo, Zhen-Dong YuanAdvances 2015 · DOI

Abstract: Based on information fusion, this research analyzes submarine cable probe positioning technology and methods. Through information classification and correlation in the submarine cable detection process, a five-parallel-type information fusion method is proposed. This approach improves information processing capacity and technical updates for submarine cable detection.

Introduction

Submarine cables constitute the critical infrastructure of global telecommunications. However, locating and maintaining these cables on the complex ocean floor remains a significant challenge. Traditional detection methods—primarily sonar, active magnetic detection, and passive magnetic detection—often operate in isolation and are heavily dependent on the subjective experience of human operators. This paper proposes a systematic, multi-layered information fusion framework designed to move submarine cable detection toward a more objective, data-driven paradigm.

The Five Parallel-Type Information Fusion Framework

The core of the proposed method is a hierarchical architecture that processes data across three distinct levels: the Data Layer, the Feature Level, and the Decision-Making Level. This structure allows for the simultaneous processing of diverse data streams, including sonar imagery, magnetic sensor readings, and navigation data.

Data Layer Fusion

The first stage of the process is the Data Layer Fusion, which serves as a critical "cleaning" phase. Raw data from various sources—such as target sea information, cable specifications, and navigation systems—are evaluated for effectiveness.

The Data Layer Fusion process involves determining data effectiveness and sampling before association.

As shown in the diagram above, data that fails the effectiveness check is discarded, while valid data is sampled and associated. This initial filtering reduces noise and ensures that only high-quality information proceeds to the more computationally intensive stages of the system.

Feature and Decision-Making Fusion

Following the data layer, the system performs Feature Level Fusion. In this phase, the system extracts specific characteristics from the cleaned data and compares them against a known feature database. This allows the system to recognize targets and filter out irrelevant environmental noise.

Finally, Decision-Making Fusion utilizes a decision library to make high-level judgments. This involves identifying the precise cable route and detecting "undesired points," such as faults or obstructions. The system generates a graphical output that assists human operators in making the final maintenance or routing decisions.

System Architecture and Feedback Loops

A defining characteristic of this research is the "five parallel-type" structure. By designing parallel processing pipelines for different data categories (e.g., environmental conditions vs. detection equipment performance), the system avoids bottlenecks and can handle large volumes of sensor data efficiently.

Comprehensive architecture of the five-parallel-type information fusion system, including the feedback mechanism.

Furthermore, the architecture incorporates a Fusion Feedback mechanism. This loop allows insights gained at the decision level to inform and optimize the data selection and processing at the lower levels in real-time, creating a more responsive and intelligent detection system.

Conclusion and Future Outlook

The proposed framework offers a qualitative leap over traditional, experience-based detection. By systematizing the fusion of sonar and magnetic data, the method improves information processing capacity and reduces the impact of artificial subjective factors.

While the theoretical framework is robust, the primary limitation remains a lack of empirical validation through field trials. Future research should focus on implementing this architecture in a real-world prototype to quantify its performance in complex deep-water environments and explore the integration of machine learning to further automate feature extraction.

Summary updated

Abstract

Based on information fusion, analyzed submarine cable probe positioning technology and methods, through the information classification and information correlation in submarine cable detection process, proposed submarine cable five parallel-type information fusion method, improved information processing capacity and technical updates of submarine cabledetection.
BibTeX
@inproceedings{zhou2015research,
  author = {Xuejun Zhou and Shuicheng Yan and Ming-jun Zuo and Zhen-Dong Yuan},
  title = {{Research on submarine cable detection technology based on information fusion}},
  booktitle = {Advances},
  year = {2015},
  doi = {10.2991/iiicec-15.2015.121},
}
Figure