Travel speed estimation using big trajectory data through an Apache Spark and Sedona based computational framework

Keywords: Big trajectory data, distributed computing, Spark
Session Type: Virtual Paper Abstract
Day: Sunday
Session Start / End Time: 2/27/2022 03:40 PM (Eastern Time (US & Canada)) - 2/27/2022 05:00 PM (Eastern Time (US & Canada))
Room: Virtual 58

Abstract

With the increasing availability of location-aware devices, passively collected big trajectory data offers researchers new opportunities for analyzing mobility and travel behaviors of drivers. This new data type offers large spatio-temporal coverage and fine spatio-temporal resolution while at the same time supporting estimates of traffic conditions. Processing big trajectory data, especially extracting information from hundreds of millions of trajectory points and assigning information to the corresponding road segments is a challenging but indispensable task for researchers who plan to make full use of the advantages of this big data for transportation analysis. In this study, we discuss a computational framework based on Apache Spark and Apache Sedona that is capable of estimating traffic speed for a state-level road network from big trajectory data. Based on the spatial object and spatial index supported by Spatial Resilient Distributed Datasets of Apache Sedona, the map-matching module in this study greatly reduces computing time while ensuring matching accuracy. We test our approach on trajectory data for the State of California and demonstrate an approach for estimating travel speed over the State’s road network. The results show that the framework proposed in this study could improve the efficiency of traffic estimation, and improve our understanding of people’s interactions with urban systems.