Paper Info
Title
Spatio-temporal joins on symbolic indoor tracking data
Abstract
To facilitate a variety of applications, positioning systems are deployed in indoor settings. For example, Bluetooth and RFID positioning are deployed in airports to support real-time monitoring of delays as well as off-line flow and space usage analyses. Such deployments generate large collections of tracking data. Like in other data management applications, joins are indispensable in this setting. However, joins on indoor tracking data call for novel techniques that take into account the limited capabilities of the positioning systems as well as the specifics of indoor spaces. This paper proposes and studies probabilistic, spatio-temporal joins on historical indoor tracking data. Two meaningful types of join are defined. They return object pairs that satisfy spatial join predicates either at a time point or during a time interval. The predicates considered include "same X," where X is a semantic region such as a room or hallway. Based on an analysis on the uncertainty inherent to indoor tracking data, effective join probabilities are formalized and evaluated for object pairs. Efficient two-phase hash-based algorithms are proposed for the point and interval joins. In a filter-and-refine framework, an R-tree variant is proposed that facilitates the retrieval of join candidates, and pruning rules are supplied that eliminate candidate pairs that do not qualify. An empirical study on both synthetic and real data shows that the proposed techniques are efficient and scalable.
Year
DOI
Venue
2011
10.1109/ICDE.2011.5767902
ICDE
Keywords
Field
DocType
object pair,rfid positioning,symbolic indoor tracking data,historical indoor tracking data,data management application,proposed technique,positioning system,indoor tracking data,indoor setting,indoor space,probabilistic logic,semantics,bluetooth,global positioning system,empirical study,radiofrequency,uncertainty,probability,data management,satisfiability,trajectory
Data mining,Joins,Computer science,Real-time computing,Global Positioning System,Hash function,Probabilistic logic,Data management,Trajectory,Bluetooth,Database,Scalability
Conference
ISSN
Citations 
PageRank 
1084-4627
14
0.66
References 
Authors
25
3
Name
Order
Citations
PageRank
Hua Lu1138083.74
Bin Yang270634.93
Christian S. Jensen3106511129.45