Comparative analysis of trajectory similarity techniques for vessels in real time : a case study on maritime traffic monitoring

Papageorgopoulos, Nikos; Παπαγεωργόπουλος, Νικόλαος

Master Thesis

Author

Papageorgopoulos, Nikos

Παπαγεωργόπουλος, Νικόλαος

Date

2023

Advisor

Doulkeridis, Christos
Δουλκερίδης, Χρήστος

View/Open

Master thesis (2.344Mb)

Keywords

Abstract

The objective of this thesis paper is to compare the performance of trajectory similarity techniques for vessels in real-time. The study presents a comprehensive review of multiple trajectory similarity techniques and identifies the most widely used methods. The methods selected for comparison include Lock-Step Euclidean distance, Dynamic Time Warping and Longest Common Subsequence. The study begins with a comprehensive examination of existing trajectory similarity techniques and their applications in the maritime domain. Following that, a fresh dataset comprised of various vessel trajectories is assembled in order to evaluate the performance and attributes of the chosen approaches. The evaluation is primarily concerned with computational efficiency. In addition to the primary focus on comparing the performance and properties of trajectory similarity techniques for real-time vessel tracking, this thesis also encompasses a thorough analysis of the tools and supplementary techniques employed in the used algorithm. A distributed processing system is employed to compute the evaluations using Spark, notably Spark Streaming for real-time data. In terms of partitioning strategies, uniform grid partitioning has been applied. Overall, a dataset of vessel trajectories is collected from a real time monitoring system of AIS, and the selected techniques are applied to the dataset. The results show that LSED performs better than the other two methods in terms of accuracy and computational efficiency. In general, the findings of this study contribute to the advancement of vessel trajectory analysis and provide guidance for selecting appropriate techniques for real-time vessel monitoring.