A new satellite-ship autonomous communication system with an integrated deep learning anomaly detection algorithm
Di Wu,Sheng Liu,Wei Wei,Yu Sui
DOI: https://doi.org/10.1007/s11042-024-18567-4
IF: 2.577
2024-02-15
Multimedia Tools and Applications
Abstract:To realize space-sea integrated communication between multiple ships under the BeiDou-3 satellite communication topology, this paper proposes a new satellite-ship autonomous communication system based on a deep learning anomaly detection algorithm. A deep learning algorithm combining coarse-grained detection (piecewise oversampling principal component analysis, POsPCA) and fine-grained sorting (VAE and differential ARIMA joint model) was established to realize two-way autonomous communication between BeiDou-3 satellites and ships. Specifically, a segmented oversampling principal component analysis algorithm was proposed to analyze anomalous BeiDou-3 short message data segments in the coarse-grained detection stage. In the fine-grained sorting stage, a joint fusion reconstruction and prediction model was proposed to calculate the fine-grained anomaly score. The autonomous communication system establishes the communication priority of multiple ships based on this score. Using the experimental platform independently built for this study to verify the algorithm performance, the proposed algorithm effectively realizes the priority ranking of anomaly scores by detecting abnormal data segments, effectively saving 64.4% of BeiDou-3 satellite short message communication resources.
computer science, information systems, theory & methods,engineering, electrical & electronic, software engineering