Cost estimation of spatial k-Nearest-neighbor operators

Ahmed M. Aly, Walid G. Aref, Mourad Ouzzani

Research output: Chapter in Book/Report/Conference proceedingConference contribution

4 Citations (Scopus)

Abstract

Advances in geo-sensing technology have led to an unprecedented spread of location-aware devices. In turn, this has resulted into a plethora of location-based services in which huge amounts of spatial data need to be efficiently consumed by spatial query processors. For a spatial query processor to properly choose among the various query processing strategies, the cost of the spatial operators has to be estimated. In this paper, we study the problem of estimating the cost of the spatial k-nearest-neighbor (k-NN, for short) operators, namely, k-NN-Select and k-NN-Join. Given a query that has a k-NN operator, the objective is to estimate the number of blocks that are going to be scanned during the processing of this operator. Estimating the cost of a k-NN operator is challenging for several reasons. For instance, the cost of a k-NN-Select operator is directly affected by the value of k, the location of the query focal point, and the distribution of the data. Hence, a cost model that captures these factors is relatively hard to realize. This paper introduces cost estimation techniques that maintain a compact set of catalog information that can be kept in main-memory to enable fast estimation via lookups. A detailed study of the performance and accuracy trade-off of each proposed technique is presented. Experimental results using real spatial datasets from OpenStreetMap demonstrate the robustness of the proposed estimation techniques.

Original languageEnglish
Title of host publicationEDBT 2015 - 18th International Conference on Extending Database Technology, Proceedings
PublisherOpenProceedings.org, University of Konstanz, University Library
Pages457-468
Number of pages12
ISBN (Electronic)9783893180677
DOIs
Publication statusPublished - 2015
Event18th International Conference on Extending Database Technology, EDBT 2015 - Brussels, Belgium
Duration: 23 Mar 201527 Mar 2015

Other

Other18th International Conference on Extending Database Technology, EDBT 2015
CountryBelgium
CityBrussels
Period23/3/1527/3/15

Fingerprint

Mathematical operators
Costs
Location based services
Query processing
Data storage equipment
Processing

ASJC Scopus subject areas

  • Information Systems
  • Software

Cite this

Aly, A. M., Aref, W. G., & Ouzzani, M. (2015). Cost estimation of spatial k-Nearest-neighbor operators. In EDBT 2015 - 18th International Conference on Extending Database Technology, Proceedings (pp. 457-468). OpenProceedings.org, University of Konstanz, University Library. https://doi.org/10.5441/002/edbt.2015.40

Cost estimation of spatial k-Nearest-neighbor operators. / Aly, Ahmed M.; Aref, Walid G.; Ouzzani, Mourad.

EDBT 2015 - 18th International Conference on Extending Database Technology, Proceedings. OpenProceedings.org, University of Konstanz, University Library, 2015. p. 457-468.

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Aly, AM, Aref, WG & Ouzzani, M 2015, Cost estimation of spatial k-Nearest-neighbor operators. in EDBT 2015 - 18th International Conference on Extending Database Technology, Proceedings. OpenProceedings.org, University of Konstanz, University Library, pp. 457-468, 18th International Conference on Extending Database Technology, EDBT 2015, Brussels, Belgium, 23/3/15. https://doi.org/10.5441/002/edbt.2015.40
Aly AM, Aref WG, Ouzzani M. Cost estimation of spatial k-Nearest-neighbor operators. In EDBT 2015 - 18th International Conference on Extending Database Technology, Proceedings. OpenProceedings.org, University of Konstanz, University Library. 2015. p. 457-468 https://doi.org/10.5441/002/edbt.2015.40
Aly, Ahmed M. ; Aref, Walid G. ; Ouzzani, Mourad. / Cost estimation of spatial k-Nearest-neighbor operators. EDBT 2015 - 18th International Conference on Extending Database Technology, Proceedings. OpenProceedings.org, University of Konstanz, University Library, 2015. pp. 457-468
@inproceedings{5beab6110c184c229d626f13d4389772,
title = "Cost estimation of spatial k-Nearest-neighbor operators",
abstract = "Advances in geo-sensing technology have led to an unprecedented spread of location-aware devices. In turn, this has resulted into a plethora of location-based services in which huge amounts of spatial data need to be efficiently consumed by spatial query processors. For a spatial query processor to properly choose among the various query processing strategies, the cost of the spatial operators has to be estimated. In this paper, we study the problem of estimating the cost of the spatial k-nearest-neighbor (k-NN, for short) operators, namely, k-NN-Select and k-NN-Join. Given a query that has a k-NN operator, the objective is to estimate the number of blocks that are going to be scanned during the processing of this operator. Estimating the cost of a k-NN operator is challenging for several reasons. For instance, the cost of a k-NN-Select operator is directly affected by the value of k, the location of the query focal point, and the distribution of the data. Hence, a cost model that captures these factors is relatively hard to realize. This paper introduces cost estimation techniques that maintain a compact set of catalog information that can be kept in main-memory to enable fast estimation via lookups. A detailed study of the performance and accuracy trade-off of each proposed technique is presented. Experimental results using real spatial datasets from OpenStreetMap demonstrate the robustness of the proposed estimation techniques.",
author = "Aly, {Ahmed M.} and Aref, {Walid G.} and Mourad Ouzzani",
year = "2015",
doi = "10.5441/002/edbt.2015.40",
language = "English",
pages = "457--468",
booktitle = "EDBT 2015 - 18th International Conference on Extending Database Technology, Proceedings",
publisher = "OpenProceedings.org, University of Konstanz, University Library",

}

TY - GEN

T1 - Cost estimation of spatial k-Nearest-neighbor operators

AU - Aly, Ahmed M.

AU - Aref, Walid G.

AU - Ouzzani, Mourad

PY - 2015

Y1 - 2015

N2 - Advances in geo-sensing technology have led to an unprecedented spread of location-aware devices. In turn, this has resulted into a plethora of location-based services in which huge amounts of spatial data need to be efficiently consumed by spatial query processors. For a spatial query processor to properly choose among the various query processing strategies, the cost of the spatial operators has to be estimated. In this paper, we study the problem of estimating the cost of the spatial k-nearest-neighbor (k-NN, for short) operators, namely, k-NN-Select and k-NN-Join. Given a query that has a k-NN operator, the objective is to estimate the number of blocks that are going to be scanned during the processing of this operator. Estimating the cost of a k-NN operator is challenging for several reasons. For instance, the cost of a k-NN-Select operator is directly affected by the value of k, the location of the query focal point, and the distribution of the data. Hence, a cost model that captures these factors is relatively hard to realize. This paper introduces cost estimation techniques that maintain a compact set of catalog information that can be kept in main-memory to enable fast estimation via lookups. A detailed study of the performance and accuracy trade-off of each proposed technique is presented. Experimental results using real spatial datasets from OpenStreetMap demonstrate the robustness of the proposed estimation techniques.

AB - Advances in geo-sensing technology have led to an unprecedented spread of location-aware devices. In turn, this has resulted into a plethora of location-based services in which huge amounts of spatial data need to be efficiently consumed by spatial query processors. For a spatial query processor to properly choose among the various query processing strategies, the cost of the spatial operators has to be estimated. In this paper, we study the problem of estimating the cost of the spatial k-nearest-neighbor (k-NN, for short) operators, namely, k-NN-Select and k-NN-Join. Given a query that has a k-NN operator, the objective is to estimate the number of blocks that are going to be scanned during the processing of this operator. Estimating the cost of a k-NN operator is challenging for several reasons. For instance, the cost of a k-NN-Select operator is directly affected by the value of k, the location of the query focal point, and the distribution of the data. Hence, a cost model that captures these factors is relatively hard to realize. This paper introduces cost estimation techniques that maintain a compact set of catalog information that can be kept in main-memory to enable fast estimation via lookups. A detailed study of the performance and accuracy trade-off of each proposed technique is presented. Experimental results using real spatial datasets from OpenStreetMap demonstrate the robustness of the proposed estimation techniques.

UR - http://www.scopus.com/inward/record.url?scp=84952815953&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84952815953&partnerID=8YFLogxK

U2 - 10.5441/002/edbt.2015.40

DO - 10.5441/002/edbt.2015.40

M3 - Conference contribution

SP - 457

EP - 468

BT - EDBT 2015 - 18th International Conference on Extending Database Technology, Proceedings

PB - OpenProceedings.org, University of Konstanz, University Library

ER -