LSH-based Similarity Query Method for Large-scale AIS Trajectory Data in Inland Waterways

This paper proposes a novel Local Sensitive Hashing (LSH)-based trajectory similarity query method specifically designed for large-scale inland Automatic Identification System (AIS) data. We first analyze the features of inland waterway AIS trajectory data, including spatial constraints, trajectory uncertainty, and irregular sampling patterns. Based on these characteristics, we develop a character-based dimensionality reduction method that transforms trajectory data into character sequences while preserving motion patterns. Our approach uses LSH techniques to enable efficient motion pattern and spatial position similarity queries on large-scale datasets. We also adopt the LCSS distance calculation for short trajectory segments to enhance query precision. Experimental results on real-world AIS data from the Yangtze River demonstrate that our method achieves higher query efficiency and accuracy than traditional approaches like DTW and Geohash-LCSS, with precision of 98%. The optimal parameters identified through extensive testing provide a balanced solution that satisfies both accuracy and efficiency requirements for practical maritime applications.