Moving objects in a geo-DBMS

1
Moving objects in a geo-DBMS

• Structuring, indexing, querying and visualizing
  moving objects in a spatiotemporal DBMS

Heraklion, Agile 2004
Marco Baars, Peter van Oosterom, Edward Verbree,
Ben Gorte
OTB Research Institute for Housing, Urban and
Mobility Studies Section GIS Technology
2
Content

• Introduction of the subject
• Generic model for moving object DMBS
• Case I static modeling
• Case II dynamic modeling
• Conclusions and recommendations

3
Introduction

• Spatiotemporal DBMSs become popular
• Traffic jams, cadastral issues
• Database is remained to stay constant
• New challenge for moving objects in database
• Databases useful for
• Large datasets
• Easy querying
• Consistency, security, redundancy,
  interoperability

4
Introduction Main question

• What is the potential and performance of the
  Oracle Spatial geo-DBMS to structure, index,
  query and visualize spatiotemporal point clouds
  of moving objects?

5
Introduction

• Vazirgiannis/Wolfson
• Map
• Moving object
• Trajectory

• Characteristic
• Developed for specific application

6
Generic model
7
Generic model

• Base table
• CREATE TABLE mov_obj
• (id, t, geometry) -prim.key id,t

8
Generic model

• Base table with (materialized) views
• CREATE VIEW move_obj_succ AS
• SELECT t1., t2.t as next_t
• FROM mov_obj t1, mov_obj t2
• WHERE t1.idt2.id and t2.t(select min(t) from
  move_obj where tgtt1.t)
• Flexible, consistent and fast

9
Case I Static modeling

• Data (id,x,y,t) collected in advance
• One base table with views in Oracle 9i Spatial
• Querying based on operators and functions
• Operator sdo_relate
• in where-clause
• Index necessary
• Function sdo_geom.relate
• Visualization (animation)

10
Case I Static modeling
11
Case II Dynamic modeling
12
Case II Dynamic modeling

• Real-time simulation
• (growing table)
• 2D and 3D
• indexing tests

13
Case II Dynamic modeling

• 2D Query SDO_RELATE
• 3D Query SDO_FILTER

x1,y1,t1
x0,y0,t0
14
Conclusions

• Generic model is flexible, fast and consistent
  for static and dynamic point data
• Choice for base table has to be made and
  depends on
• 2D or 3D queries
• 2D or 3D index
• Update time for index depends on covered area and
  number of objects in growing table

15
Future Research

• Prove Generic Model for 4D data (x,y,z,t)
• Test the ArcIMS Tracking Server
• Implementing more efficient R-tree
• Test model for polylines, polygons or polyhedrons