Analysis of a Research Paper

1
Analysis of a Research Paper

• Tom Whitfield
• David Van Riper
• Csci 8715 Group 3

2
Outline

• Introduction to the problem
• Complexity of spatio-temporal queries
• Sharing as a method to deal with these
  complexities
• Other considerations
• Conclusion

3
Locating Moving Objects

• Hindrances of efficient queries
• Scalability
• Large number of objects
• Large number of queries
• Complexity
• Intensive algorithms
• CPU, I/O intensive
• Scalability yet to be addressed

4
S-T Query Classification Time

• Historical tracking of past events
• Ex Locations of one object in the past hour
• NOW current location of a moving object
• Ex What is the nearest gas station?
• Future Moving object prediction
• Ex Tell me when a particular object enters a
  specific region

5
S-T Query Classification Mutability

• Stationary Queries on Moving Objects
• Ex Find the 10 nearest taxis to Mariucci
• Moving Queries on Stationary Objects
• Ex As I am moving, find the nearest ATM
• Moving Queries on Moving Objects
• Ex As I am moving, find all taxis within a
  one-mile radius

6
Sharing

• provide a scalable framework to support any
  kind of concurrent continuous spatio-temporal
  queries
• Preprocessing of queries
• Continuous query
• Query parser
• Query optimizer
• Separate groups of queries

7
Sharing

• Multiple concurrent queries with common
  sub-expressions
• However
• Not scalable (limited number of queries)
• Should be known a priori
• Only deals with snapshot queries
• Framework Extend to support continuous
  spatio-temporal queries

8
Sharing Examples

• Space
• Spatial hierarchies based on area, overlap
• Cost models based on containment
• Decision tree
• Query Operator
• Perform joins between spatial and non-spatial
  attributes
• Cost of indexing should not be excessive

9
Sharing Examples

• Object Interest
• Joins after common interests determined
• Selection
• Perform filters before the join operations
• Eliminate extraneous instances
• Window Join
• Use same sliding window in multiple queries

10
Other Considerations

• Difficult to maintain spatial indices
• How well can data be accurately represented?
• Non-database considerations
• Assume GPS availability improves for non-E911
  applications
• Effectively balance societal benefits with
  detractions for public implementation

11
Conclusion

• Little research has been done in this area thus
  far
• Many different approaches to make operations more
  efficient
• Developing technology substantiating need for
  spatio-temporal database optimization
• Need to look beyond computation