Ship Trajectory Outlier Detection Service System Based on Collaborative Computing

In order to ensure the safety of ships during the voyage, we need to use the AIS data to find outlying ship trajectories and remind other ships to take the necessary avoidance actions. In the process of ship trajectory outlier detection, on the one hand, the ship trajectory outlier detection model trained on historical data is needed, on the other hand, the requirement for real-time detection should be met. Therefore, this paper designs ship trajectory outlier detection service system based on collaborative computing. The service system can combine the advantages of batch computing framework and stream computing framework. Trajectory data services, real-time annotation service are implemented in stream computing framework, F-DBSCAN outlier detection service, model training service, and model-based outlier detection service are implemented in batch computing framework. Memory database is used to complete data interaction between the two frameworks. The experiment shows that the service system can detect the outlying ship trajectories according to the real-time AIS data while using the outlier detection model.