Technical Report: Towards Efficient Indexing of Spatiotemporal Trajectories on the GPU for Distance Threshold Similarity Searches
Michael Gowanlock, Henri Casanova

TL;DR
This paper introduces three GPU-based indexing strategies for spatiotemporal trajectory similarity searches, demonstrating significant performance improvements over CPU implementations in various scenarios.
Contribution
It develops novel GPU-specific indexing methods for trajectory searches and analyzes their performance conditions, advancing trajectory processing efficiency.
Findings
GPU indexes outperform multithreaded CPU indexes in experiments
Performance depends on spatial, temporal, and spatiotemporal selectivity
GPU methods are effective for real-world and synthetic datasets
Abstract
Applications in many domains require processing moving object trajectories. In this work, we focus on a trajectory similarity search that finds all trajectories within a given distance of a query trajectory over a time interval, which we call the distance threshold similarity search. We develop three indexing strategies with spatial, temporal and spatiotemporal selectivity for the GPU that differ significantly from indexes suitable for the CPU, and show the conditions under which each index achieves good performance. Furthermore, we show that the GPU implementations outperform multithreaded CPU implementations in a range of experimental scenarios, making the GPU an attractive technology for processing moving object trajectories. We test our implementations on two synthetic and one real-world dataset of a galaxy merger.
Peer Reviews
No public reviews on file for this paper yet. If you reviewed it on a platform where reviews are public (OpenReview, ICLR, NeurIPS, ICML), you can paste yours below so the community can read it here.
Videos
No videos yet. Explain this paper in a talk, walkthrough, or lecture? Add one.
Taxonomy
TopicsData Management and Algorithms · Geographic Information Systems Studies · Advanced Image and Video Retrieval Techniques
