Yun Wen

24
FYTG 5101 Presentation Data Management For Spatio-temporal Data Of PM 2.5 Indices Through Mobile Sensing Type of This Project: Survey Yun Wen

description

FYTG 5101 Presentation Data Management For Spatio -temporal Data Of PM 2.5 Indices Through Mobile Sensing Type of This Project: Survey. Yun Wen. Outlines. Motivation ST Concepts and Models Storage Structure and Indexing. Reliable?. Project: Home Air Quality Sensor Network. CO, O 2 , - PowerPoint PPT Presentation

Transcript of Yun Wen

Page 1: Yun Wen

FYTG 5101 Presentation

Data Management For Spatio-temporal Data Of PM 2.5 Indices Through Mobile SensingType of This Project: Survey

Yun Wen

Page 2: Yun Wen

Outlines

• Motivation• ST Concepts and Models• Storage Structure and Indexing

Page 3: Yun Wen
Page 4: Yun Wen
Page 5: Yun Wen

Reliable?

Page 6: Yun Wen

Project: Home Air Quality Sensor Network

CO,O2,PM2.5Smoke,TempHumidity

+Electricity Consumption

Page 7: Yun Wen

Spatiotemporal Data

Page 8: Yun Wen

Spatiotemporal Data

Page 9: Yun Wen

SpatioTemporal DataBase Management System

• Temporal DBMS: existance time, valid-time, transaction-time

• Spatial DBMS: GIS, environmental information system, trajectory and multimedia

Page 10: Yun Wen

ER Temporal

Page 11: Yun Wen

ER Spatial

Page 12: Yun Wen

ER SpatioTemporal

Page 13: Yun Wen

Storage Structure and Indexing

1D database indexing structures are not appropriate• Quad trees• K-D-B trees• R-tree• …..

Page 14: Yun Wen

Quad tree

Page 15: Yun Wen

Quad tree

Type of Data:• region quadtree• point quadtree• Lines/curves quadtreeNode Information:• Four pointers

(NW,NE,SW,SE)• Point: key(x,y);value

Page 16: Yun Wen

Quad tree

Do not take paging of secondary memory into accountTemporal:Linear Quadtree

Improved: Overlapping Linear Quadtree

Page 17: Yun Wen

K-D-B tree

Page 18: Yun Wen

K-D-B tree

Region pages (region,child) + point pages(point,location)

• Same path length • Disjoint page• Union of all region in the

child is region

Page 19: Yun Wen

K-D-B tree

Designed for paged memory, but are useful only for point data

Page 20: Yun Wen

B+ tree

Page 21: Yun Wen

R tree

• A B+ tree

Page 22: Yun Wen

R tree

Page 23: Yun Wen

R tree

Balanced-search treeOrganize data in pageOverlapped regionn entries M maximum

Variants:Hilbert R-treeX-treeR* tree, R+ tree

Page 24: Yun Wen

Conclusion

• All trees are important• Trade-off