Querying Imprecise Data In Moving Object Environments

1
Querying Imprecise Data In Moving Object
Environments

• Reynold Cheng, Sunil Prabhakar, Dmitri V.
  Kalashnikov
• Department of Computer Sciences, Purdue
  University
• Email ckcheng, sunil, dvk_at_cs.purdue.edu
• Pervasive Location Aware Computing Environments
  (PLACE) http//www.cs.purdue.edu/place

2
Data Imprecision Problem

• Objects move continuously
• Limited bandwidth and battery power
• RESULT The database cannot keep track of the
  actual locations
• Use old database values ? incorrect results
• Can provide guarantees if object locations are
  bounded.

3
Problem Definition

• Use maximum velocity and/or possible routes to
  derive uncertainty region
• Object is located in the uncertainty region with
  a probability density function
• A generic algorithm for querying uncertainty
  regions with general shapes
• Augment query answers with probabilities

C
A
B
D
4
Probabilistic Range Query (PRQ)

• Let Ai overlap of uncertainty region of object
  i and R
• Pi ?Aifi(x,y) dxdy
• If Pi ? 0, return (i,Pi)

R
5
Probabilistic Nearest Neighbor Query (PNNQ)

• At distance r, A is the nearest neighbor of Q if
• A is at distance r from Q
• B,C,D are all located at distances gt r from Q.
• The pdf pA(r) can be computed.

C
A
r
Q
D
B
6
Probabilistic Nearest Neighbor Query (PNNQ)

• Compute pA(r)
• From the shortest distance of A to Q (nA)
• To the longest distance of A to Q (fA)

C
fA
A
nA
Q
D
B