Fast trajectory search for real-world applications

Wang, S 2019, Fast trajectory search for real-world applications, Doctor of Philosophy (PhD), Science, RMIT University.

Document type: Thesis
Collection: Theses

Attached Files
Name Description MIMEType Size
Wang.pdf Thesis application/pdf 11.55MB
Title Fast trajectory search for real-world applications
Author(s) Wang, S
Year 2019
Abstract With the popularity of smartphones equipped with GPS, a vast amount of trajectory data are being produced from location-based services, such as Uber, Google Maps, and Foursquare. We broadly divide trajectory data into three types: 1) commuter trajectories from taxicabs and ride-sharing apps; 2) vehicle trajectories from GPS navigation apps; 3) activity trajectories from social network check-ins and travel blogs.

We investigate efficient and effective search on each of the three types of trajectory data, each of which has a real-world application. In particular: 1) commuter trajectory search can serve for the transport capacity estimation and route planning; 2) vehicle trajectory search can help real-time traffic monitoring and trend analysis; 3) activity trajectory search can be used in interactive and personalized trip planning.

As the most straightforward trajectory data, a commuter trajectory only contains two points: origin and destination indicating a passenger’s movement, which is valuable for transportation decision making. In this thesis, we propose a novel query RkNNT to estimate the capacity of a bus route in the transport network. Answering RkNNT is challenging due to the high amount of data from commuters. We propose efficient solutions to prune most trajectories which cannot choose a query route as their nearest one. Further, we apply RkNNT to the optimal route planning problem-MaxRkNNT.

A vehicle trajectory has more points than a commuter trajectory, as it tracks the whole trace of a vehicle and can further advocate the application of traffic monitoring. We conclude the common queries over trajectory data for monitoring purposes and proposes a search engine Torch to manage and search trajectories with map matching over a road network, instead of storing raw data sampled from GPS with a high cost. Besides improving the efficiency of search, Torch also supports compression, effectiveness evaluation of various existing similarity measures, and large-scale clustering k-paths with a novel similarity measure LORS.

Exploring the activity trajectory data which contains textual information can help plan personalized trips for tourists. Based on spatial indexes which we propose for commuter and vehicle trajectory data, we further develop a unified search paradigm to process various top-k queries over activity trajectory and POIs data (hotels, restaurants, and attractions, etc.) at the same time. In particular, a new point-wise similarity measure PATS and an indexing framework with a unified search paradigm are proposed.
Degree Doctor of Philosophy (PhD)
Institution RMIT University
School, Department or Centre Science
Subjects Database Management
Keyword(s) trajectory data
spatial database
traffic monitoring
trip planning
Note © 2019 Sheng Wang. All Rights Reserved.
Version Filter Type
Access Statistics: 267 Abstract Views, 216 File Downloads  -  Detailed Statistics
Created: Wed, 14 Aug 2019, 09:45:23 EST by Keely Chapman
© 2014 RMIT Research Repository • Powered by Fez SoftwareContact us