Open AccessMIDET: A Method for Indexing Traffic Events
Author(s) -
Mariana M. Garcez Duarte,
Marcos V. Pontarolo,
Rebeca Schroeder,
Carmem S. Hara
Publication year - 2021
Language(s) - English
Resource type - Conference proceedings
DOI - 10.5753/sbbd.2021.17879
Subject(s) - computer science , search engine indexing , joins , data mining , grid , skew , bitmap , set (abstract data type) , tree (set theory) , data structure , real time computing , information retrieval , artificial intelligence , geometry , telecommunications , mathematical analysis , mathematics , programming language
Traffic events announcements such as jams and road closures are continuously reported by mobile and Web applications. This collection of spatio-temporal data is an important source of information for urban planning, and can be used to orchestrate a number of actions to mprove the mobility, such as traffic control, traffic lights synchronization and preventive maintenance. Such analysis usually involves computation of spatial relationships among data, and may involve location of landmarks, roads and different types of events. In this paper, we propose a Method for Indexing Traffic Events (MIDET) for querying spatio-temporal data, whose location can be represented as a point or collection of points. MIDET is based on a fixed-grid space-oriented partitioning. In order to tackle the data skew, each grid cell is associated with a set of blocks containing event records. Moreover, a bitmap index is used for filtering out blocks without retrieving the actual data. MIDET provides the following benefits: adoption of a simple bulk loading process to manage dynamic insertion streams, and in-memory spatial joins. We conducted an experimental study using real data obtained from Waze. MIDET’s query performance was compared with Postgis, which adopts an R-tree index structure.