Apr 8, 2011 - c8 c9 c7 c6 c5 c4 c3 c2. Facilities (F). Clients (C). Xiaokui Xiao, Bin Yao, Feifei Li. Optimal Location Queries in Road Network Databases ...
Optimal Location Queries in Road Network Databases Xiaokui Xiao1 , Bin Yao2 , Feifei Li2
1 School of Computer Engineering Nanyang Technological University, Singapore
April 8, 2011
2
Department of Computer Science Florida State University, USA
Introduction and Motivation An optimal location (OL) query:
Xiaokui Xiao, Bin Yao, Feifei Li
Optimal Location Queries in Road Network Databases
Introduction and Motivation An optimal location (OL) query: Facilities (F)
f3
f1
f2
Xiaokui Xiao, Bin Yao, Feifei Li
Optimal Location Queries in Road Network Databases
Introduction and Motivation An optimal location (OL) query:
c7
c9
c6
c8
c1 f1
f3
c2
c3
f2
Xiaokui Xiao, Bin Yao, Feifei Li
Facilities (F) Clients (C)
c5 c4
Optimal Location Queries in Road Network Databases
Introduction and Motivation An optimal location (OL) query:
c7
c9
c6
c8
c1 f1
f3
c2
c3
f2
Xiaokui Xiao, Bin Yao, Feifei Li
Facilities (F) Clients (C)
c5
Candidates (P)
c4
Optimal Location Queries in Road Network Databases
Introduction and Motivation An optimal location (OL) query:
v6
c7
c9 c1
f3 v5
c5
c6
c8 v2 c 3 f1 v c 2 1
f2
Xiaokui Xiao, Bin Yao, Feifei Li
Facilities (F) Clients (C) Candidates (P)
v4 v3
c4
Optimal Location Queries in Road Network Databases
Introduction and Motivation An optimal location (OL) query:
v6
c7
c9 c1
f3 v5
c5
c6
c8 v2 c 3 f1 v c 2 1
f2
Facilities (F) Clients (C) Candidates (P)
v4 v3
c4
a(c3)=d (c3, f2)
Xiaokui Xiao, Bin Yao, Feifei Li
Optimal Location Queries in Road Network Databases
Introduction and Motivation An optimal location (OL) query:
v6
c7
c9 c1
f3 v5
c5
c6
c8 v2 c 3 f1 v c 2 1
f2
Facilities (F) Clients (C) Candidates (P)
v4 v3
c4
a(c3)=d (c3, f2)
Xiaokui Xiao, Bin Yao, Feifei Li
Optimal Location Queries in Road Network Databases
Introduction and Motivation An optimal location (OL) query:
v6
c7
c9 c1
f3 v5
c5
c6
c8 v2 c 3 f1 v c 2 1
f2
Facilities (F) Clients (C) Candidates (P)
v4 v3
c4
a(c3)=d (c3, f2)
Xiaokui Xiao, Bin Yao, Feifei Li
Optimal Location Queries in Road Network Databases
Introduction and Motivation An optimal location (OL) query:
v6
c7
c9 c1
f3 v5
c5
c6
c8 v2 c 3 f1 v c 2 1
f2
Facilities (F) Clients (C) Candidates (P)
v4 v3
c4
a(c3)=d (c3, f2)
Cabello et al. [1] and Wong et al. [2] deal with competitive location queries in the L2 space
[1] S. Cabello, et al. Reverse facility location problems. In CCCG, 2005 [2] R. Wong, et al. Efficient method for maximizing bichromatic reverse nearest neighbor. In PVLDB, 2009
Xiaokui Xiao, Bin Yao, Feifei Li
Optimal Location Queries in Road Network Databases
Introduction and Motivation An optimal location (OL) query:
v6
c7
c9 c1
f3 v5
c5
c6
c8 v2 c 3 f1 v c 2 1
f2
Facilities (F) Clients (C) Candidates (P)
v4 v3
c4
a(c3)=d (c3, f2)
Cabello et al. [1] and Wong et al. [2] deal with competitive location queries in the L2 space Du et al. [3] and Zhang et al. [4] investigate competitive and MinSum location queries in the L1 space, respectively. [1] S. Cabello, et al. Reverse facility location problems. In CCCG, 2005 [2] R. Wong, et al. Efficient method for maximizing bichromatic reverse nearest neighbor. In PVLDB, 2009 [3] Y. Du, et al. The optimal-location query. In SSTD, 2005 [4] D. Zhang, et al. Progressive computation of the min-dist optimal-location query. In VLDB, 2006
Xiaokui Xiao, Bin Yao, Feifei Li
Optimal Location Queries in Road Network Databases
Problem Formulation
Competitive location query: p
=
argmax |Cp |, p∈P
where Cp is the set of clients attracted by p.
Xiaokui Xiao, Bin Yao, Feifei Li
Optimal Location Queries in Road Network Databases
Problem Formulation
Competitive location query: p
=
argmax |Cp |, p∈P
where Cp is the set of clients attracted by p.
v6
c7
c9 c1
c8
f3 v5
c5
c6
v2 c c 3 f f1 v 2 2 1 Xiaokui Xiao, Bin Yao, Feifei Li
Facilities (F) Clients (C) Candidates (P)
v4 v3
c4
Optimal Location Queries in Road Network Databases
Problem Formulation
Competitive location query: p
=
argmax |Cp |, p∈P
where Cp is the set of clients attracted by p.
Example 1: Given existing supermarkets F (residential locations C ) in a city, Julie want to open a new supermarket that can attract as many customers as possible.
Xiaokui Xiao, Bin Yao, Feifei Li
Optimal Location Queries in Road Network Databases
Problem Formulation
MinSum location query: p
=
argmin p∈P
Xiaokui Xiao, Bin Yao, Feifei Li
X
a(c).
c∈C
Optimal Location Queries in Road Network Databases
Problem Formulation
MinSum location query: p
=
argmin p∈P
X
a(c).
c∈C
Example 2: John owns a set F of pizza shops that deliver to a set C of places in a city. He looks for a location to add another pizza shop to minimize the average distance from the place in C to their respective nearest shops.
Xiaokui Xiao, Bin Yao, Feifei Li
Optimal Location Queries in Road Network Databases
Problem Formulation
MinMax location query: p
=
argmin max a(c) . p∈P
Xiaokui Xiao, Bin Yao, Feifei Li
c∈C
Optimal Location Queries in Road Network Databases
Problem Formulation
MinMax location query: p
=
argmin max a(c) . p∈P
c∈C
Example 3: Given the set F (C ) of existing fire stations (buildings) in a city, the government may seek a candidate location that minimizes the maximum distance from any building to its nearest fire station.
Xiaokui Xiao, Bin Yao, Feifei Li
Optimal Location Queries in Road Network Databases
Solution Overview v6
c7
c9 c1 f1
f3 v5 c6
c8 v1
v c2 2 c3
f2
Xiaokui Xiao, Bin Yao, Feifei Li
Facilities (F) Clients (C)
c5 v4 v3
Candidates (P)
c4
Optimal Location Queries in Road Network Databases
Solution Overview v6
c9 c1 f1
v6
c7
v1
v c2 2 c3
c7
v4 v3
f3 v5
v c2 2 c3
f2
Facilities (F) Clients (C)
v4 v3
Candidates (P)
c4
c5
c6
c8 v1
f2
Facilities (F) Clients (C)
c5
c6
c8
c9 c1 f1
f3 v5
Candidates (Ec )
c4
Construct the road intervals.
Xiaokui Xiao, Bin Yao, Feifei Li
Optimal Location Queries in Road Network Databases
Solution Overview v6
c9 c1 f1
v6
c7
v1
v c2 2 c3
c7
v4 v3
f3 v5
v c2 2 c3
f2
Facilities (F) Clients (C)
v4 v3
Candidates (P)
c4
c5
c6
c8 v1
f2
Facilities (F) Clients (C)
c5
c6
c8
c9 c1 f1
f3 v5
Candidates (Ec )
c4
Construct the road intervals. Traverse the candidate road intervals in a certain order.
Xiaokui Xiao, Bin Yao, Feifei Li
Optimal Location Queries in Road Network Databases
Solution Overview v6
c9 c1 f1
v6
c7
v1
v c2 2 c3
c7
v4
f3 v5
v c2 2 c3
f2
Candidates (P)
c4
v3
Facilities (F) Clients (C)
c5
c6
c8 v1
f2
Facilities (F) Clients (C)
c5
c6
c8
c9 c1 f1
f3 v5
v4
Candidates (Ec )
c4
v3
Construct the road intervals. Traverse the candidate road intervals in a certain order. Identify the local optimal locations.
Xiaokui Xiao, Bin Yao, Feifei Li
Optimal Location Queries in Road Network Databases
Solution Overview v6
c9 c1 f1
v6
c7
v1
v c2 2 c3
c7
v4 v3
f3 v5
v c2 2 c3
f2
Candidates (P)
c4
Facilities (F) Clients (C)
c5
c6
c8 v1
f2
Facilities (F) Clients (C)
c5
c6
c8
c9 c1 f1
f3 v5
v4
Candidates (Ec )
c4
v3
Construct the road intervals. Traverse the candidate road intervals in a certain order. Identify the local optimal locations. Return the global optimal locations. Xiaokui Xiao, Bin Yao, Feifei Li
Optimal Location Queries in Road Network Databases
Local optimal locations: competitive location queries
Xiaokui Xiao, Bin Yao, Feifei Li
Optimal Location Queries in Road Network Databases
Local optimal locations: competitive location queries Lemma A client c is attracted by a point p on an edge e ∈ Ec , iff there exists an entry hc, d(c, v )i in the attraction set of an endpoint v of e, such that d(c, v ) + d(v , p) ≤ a(c).
Xiaokui Xiao, Bin Yao, Feifei Li
Optimal Location Queries in Road Network Databases
Local optimal locations: competitive location queries Lemma A client c is attracted by a point p on an edge e ∈ Ec , iff there exists an entry hc, d(c, v )i in the attraction set of an endpoint v of e, such that d(c, v ) + d(v , p) ≤ a(c).
vl
vr
d(c, vl ) = 4
c
f a(c) = 5
Xiaokui Xiao, Bin Yao, Feifei Li
Optimal Location Queries in Road Network Databases
Local optimal locations: competitive location queries Lemma A client c is attracted by a point p on an edge e ∈ Ec , iff there exists an entry hc, d(c, v )i in the attraction set of an endpoint v of e, such that d(c, v ) + d(v , p) ≤ a(c).
a(c) − d(c, vl ) = 1
vl
vr
d(c, vl ) = 4
c
f a(c) = 5
Xiaokui Xiao, Bin Yao, Feifei Li
Optimal Location Queries in Road Network Databases
Local optimal locations: competitive location queries Lemma A client c is attracted by a point p on an edge e ∈ Ec , iff there exists an entry hc, d(c, v )i in the attraction set of an endpoint v of e, such that d(c, v ) + d(v , p) ≤ a(c).
A(vl )
A(vr )
< c1 , 4 >
< c2 , 3 >
< c3 , 1 > < c4 , 3 >
< c3 , 2 > < c4 , 4 >
vl
e(length = 5) a(ci ) = 5
vr
Xiaokui Xiao, Bin Yao, Feifei Li
Optimal Location Queries in Road Network Databases
Local optimal locations: competitive location queries Lemma A client c is attracted by a point p on an edge e ∈ Ec , iff there exists an entry hc, d(c, v )i in the attraction set of an endpoint v of e, such that d(c, v ) + d(v , p) ≤ a(c).
A(vl )
A(vr )
< c1 , 4 >
< c2 , 3 >
< c3 , 1 > < c4 , 3 >
< c3 , 2 > < c4 , 4 >
vl
e(length = 5) a(ci ) = 5
vr
Xiaokui Xiao, Bin Yao, Feifei Li
0
1
2
3
4
5 R
Optimal Location Queries in Road Network Databases
Local optimal locations: competitive location queries Lemma A client c is attracted by a point p on an edge e ∈ Ec , iff there exists an entry hc, d(c, v )i in the attraction set of an endpoint v of e, such that d(c, v ) + d(v , p) ≤ a(c).
A(vl )
A(vr )
< c1 , 4 >
< c2 , 3 >
< c3 , 1 > < c4 , 3 >
< c3 , 2 > < c4 , 4 >
vl
e(length = 5) a(ci ) = 5
vr
Xiaokui Xiao, Bin Yao, Feifei Li
0
1
2
3
4
5 R
Optimal Location Queries in Road Network Databases
Local optimal locations: competitive location queries Lemma A client c is attracted by a point p on an edge e ∈ Ec , iff there exists an entry hc, d(c, v )i in the attraction set of an endpoint v of e, such that d(c, v ) + d(v , p) ≤ a(c).
A(vl )
A(vr )
< c1 , 4 >
< c2 , 3 >
< c3 , 1 > < c4 , 3 >
< c3 , 2 > < c4 , 4 >
vl
e(length = 5) a(ci ) = 5
vr
Xiaokui Xiao, Bin Yao, Feifei Li
s1
0
1
2
3
4
5 R
Optimal Location Queries in Road Network Databases
Local optimal locations: competitive location queries Lemma A client c is attracted by a point p on an edge e ∈ Ec , iff there exists an entry hc, d(c, v )i in the attraction set of an endpoint v of e, such that d(c, v ) + d(v , p) ≤ a(c).
A(vl )
A(vr )
< c1 , 4 >
< c2 , 3 >
< c3 , 1 > < c4 , 3 >
< c3 , 2 > < c4 , 4 >
vl
e(length = 5) a(ci ) = 5
vr
Xiaokui Xiao, Bin Yao, Feifei Li
s1
0
1
2
3
4
5 R
Optimal Location Queries in Road Network Databases
Local optimal locations: competitive location queries Lemma A client c is attracted by a point p on an edge e ∈ Ec , iff there exists an entry hc, d(c, v )i in the attraction set of an endpoint v of e, such that d(c, v ) + d(v , p) ≤ a(c).
A(vl )
A(vr )
< c1 , 4 >
< c2 , 3 >
< c3 , 1 > < c4 , 3 >
< c3 , 2 > < c4 , 4 >
vl
e(length = 5) a(ci ) = 5
vr
Xiaokui Xiao, Bin Yao, Feifei Li
s1 s2
0
1
2
3
4
5 R
Optimal Location Queries in Road Network Databases
Local optimal locations: competitive location queries Lemma A client c is attracted by a point p on an edge e ∈ Ec , iff there exists an entry hc, d(c, v )i in the attraction set of an endpoint v of e, such that d(c, v ) + d(v , p) ≤ a(c).
A(vl )
A(vr )
< c1 , 4 >
< c2 , 3 >
< c3 , 1 > < c4 , 3 >
< c3 , 2 > < c4 , 4 >
vl
e(length = 5) a(ci ) = 5
vr
Xiaokui Xiao, Bin Yao, Feifei Li
s1 s2 s3 s4
s40
0
1
2
3
4
5 R
Optimal Location Queries in Road Network Databases
Local optimal locations: competitive location queries Lemma A client c is attracted by a point p on an edge e ∈ Ec , iff there exists an entry hc, d(c, v )i in the attraction set of an endpoint v of e, such that d(c, v ) + d(v , p) ≤ a(c).
A(vl )
A(vr )
< c1 , 4 >
< c2 , 3 >
< c3 , 1 > < c4 , 3 >
< c3 , 2 > < c4 , 4 >
vl
e(length = 5) a(ci ) = 5
vr
Xiaokui Xiao, Bin Yao, Feifei Li
s1 s2 s3 s4
s40
0
1
2
3
4
5 R
Optimal Location Queries in Road Network Databases
Computing attractor distances and attraction sets
Xiaokui Xiao, Bin Yao, Feifei Li
Optimal Location Queries in Road Network Databases
Computing attractor distances and attraction sets Computing attractor distances: Erwig and Hagen’s algorithm.
Xiaokui Xiao, Bin Yao, Feifei Li
Optimal Location Queries in Road Network Databases
Computing attractor distances and attraction sets Computing attractor distances: Erwig and Hagen’s algorithm.
v6
c7
c9 c1 f1
c5
c6
c8 v1
f3 v5
v c2 2 c3
f2
Xiaokui Xiao, Bin Yao, Feifei Li
v4 v3
c4
Optimal Location Queries in Road Network Databases
Computing attractor distances and attraction sets Computing attractor distances: Erwig and Hagen’s algorithm.
v6
c7
c9 c1 f1
c5
c6
c8 v1
f3 v5
v c2 2 c3
f2
Xiaokui Xiao, Bin Yao, Feifei Li
v4 v3
c4
Optimal Location Queries in Road Network Databases
Computing attractor distances and attraction sets Computing attractor distances: Erwig and Hagen’s algorithm.
v6
c7
c9 c1 f1
c5
c6
c8 v1
f3 v5
v c2 2 c3
f2
Xiaokui Xiao, Bin Yao, Feifei Li
v4 v3
c4
Optimal Location Queries in Road Network Databases
Computing attractor distances and attraction sets Computing attractor distances: Erwig and Hagen’s algorithm.
v6
c7
c9 c1 f1
c5
c6
c8 v1
f3 v5
v c2 2 c3
f2
Xiaokui Xiao, Bin Yao, Feifei Li
v4 v3
c4
Optimal Location Queries in Road Network Databases
Computing attractor distances and attraction sets Computing attractor distances: Erwig and Hagen’s algorithm.
v6
c7
c9 c1 f1
c5
c6
c8 v1
f3 v5
v c2 2 c3
f2
v4 v3
c4
Computing attraction sets: The Blossom and OTF algorithms.
Xiaokui Xiao, Bin Yao, Feifei Li
Optimal Location Queries in Road Network Databases
Computing attractor distances and attraction sets Computing attractor distances: Erwig and Hagen’s algorithm.
v6
c7
c9 c1 f1
c5
c6
c8 v1
f3 v5
v c2 2 c3
f2
v4 v3
c4
Computing attraction sets: The Blossom and OTF algorithms. The Blossom algorithm, Time: O(n2 log n), space: O(n2 ).
Xiaokui Xiao, Bin Yao, Feifei Li
Optimal Location Queries in Road Network Databases
Computing attractor distances and attraction sets Computing attractor distances: Erwig and Hagen’s algorithm.
v6
c7
c9 c1 f1
c5
c6
c8 v1
f3 v5
v c2 2 c3
f2
v4 v3
c4
Computing attraction sets: The Blossom and OTF algorithms. The Blossom algorithm, Time: O(n2 log n), space: O(n2 ). The OTF algorithm
Xiaokui Xiao, Bin Yao, Feifei Li
Optimal Location Queries in Road Network Databases
The OTF algorithm: construct the A(v ) on the fly
Xiaokui Xiao, Bin Yao, Feifei Li
Optimal Location Queries in Road Network Databases
The OTF algorithm: construct the A(v ) on the fly A straightforward solution: apply Dijkstra’s algorithm to scan all vertices starting at v . If d(v , c) < a(c), add c into A(v ).
Xiaokui Xiao, Bin Yao, Feifei Li
Optimal Location Queries in Road Network Databases
The OTF algorithm: construct the A(v ) on the fly A straightforward solution: apply Dijkstra’s algorithm to scan all vertices starting at v . If d(v , c) < a(c), add c into A(v ).
Lemma Given two vertices v and v 0 in G , such that d(v , v 0 ) is larger than the distance from v 0 to its nearest facility f 0 . Then, ∀c ∈ A(v ), the shortest path from v to c must not go through v 0 .
Xiaokui Xiao, Bin Yao, Feifei Li
Optimal Location Queries in Road Network Databases
The OTF algorithm: construct the A(v ) on the fly A straightforward solution: apply Dijkstra’s algorithm to scan all vertices starting at v . If d(v , c) < a(c), add c into A(v ).
Lemma Given two vertices v and v 0 in G , such that d(v , v 0 ) is larger than the distance from v 0 to its nearest facility f 0 . Then, ∀c ∈ A(v ), the shortest path from v to c must not go through v 0 .
G1
f0 1
2 v
v0
Xiaokui Xiao, Bin Yao, Feifei Li
G2 G3
Optimal Location Queries in Road Network Databases
The OTF algorithm: construct the A(v ) on the fly A straightforward solution: apply Dijkstra’s algorithm to scan all vertices starting at v . If d(v , c) < a(c), add c into A(v ).
Lemma Given two vertices v and v 0 in G , such that d(v , v 0 ) is larger than the distance from v 0 to its nearest facility f 0 . Then, ∀c ∈ A(v ), the shortest path from v to c must not go through v 0 .
G1
f0 1
2 v
A(v )
v0
G2 G3
< c2, 1.2 >
..
Xiaokui Xiao, Bin Yao, Feifei Li
Optimal Location Queries in Road Network Databases
The OTF algorithm: construct the A(v ) on the fly A straightforward solution: apply Dijkstra’s algorithm to scan all vertices starting at v . If d(v , c) < a(c), add c into A(v ).
Lemma Given two vertices v and v 0 in G , such that d(v , v 0 ) is larger than the distance from v 0 to its nearest facility f 0 . Then, ∀c ∈ A(v ), the shortest path from v to c must not go through v 0 .
G1
f0 1
2 v
A(v )
v0
G2 G3
< c2, 1.2 >
..
Xiaokui Xiao, Bin Yao, Feifei Li
Optimal Location Queries in Road Network Databases
The OTF algorithm: construct the A(v ) on the fly
v6
c7
c9 c1 f1
c5
c6
c8 v1
f3 v5
v c2 2 c3
Xiaokui Xiao, Bin Yao, Feifei Li
f2
v4 v3
c4
Optimal Location Queries in Road Network Databases
The OTF algorithm: construct the A(v ) on the fly
v6
c7
c9 c1 f1
c5
c6
c8 v1
f3 v5
v c2 2 c3
f2
v4 v3
c4 A(v2)
Xiaokui Xiao, Bin Yao, Feifei Li
Optimal Location Queries in Road Network Databases
The OTF algorithm: construct the A(v ) on the fly
v6
c7
c9 c1 f1
c5
c6
c8 v1
f3 v5
v c2 2 c3
d (v2, c2) ≤ a(c2)
Xiaokui Xiao, Bin Yao, Feifei Li
f2
v4 v3
c4 A(v2)
Optimal Location Queries in Road Network Databases
The OTF algorithm: construct the A(v ) on the fly
v6
c7
c9 c1 f1
c5
c6
c8 v1
f3 v5
v c2 2 c3
d (v2, c2) ≤ a(c2)
Xiaokui Xiao, Bin Yao, Feifei Li
f2
v4 v3
c4
A(v2) < c2 , 1 >
Optimal Location Queries in Road Network Databases
The OTF algorithm: construct the A(v ) on the fly
v6
c7
c9 c1 f1
c5
c6
c8 v1
f3 v5
v c2 2 c3
d (v2, v1) > d (v1, f1)
Xiaokui Xiao, Bin Yao, Feifei Li
f2
v4 v3
c4
A(v2) < c2 , 1 >
Optimal Location Queries in Road Network Databases
The OTF algorithm: construct the A(v ) on the fly
v6
c7
c9 c1 f1
c5
c6
c8 v1
f3 v5
v c2 2 c3
f2
v4 v3
c4
A(v2) < c2 , 1 > < c3 , 2 >
Xiaokui Xiao, Bin Yao, Feifei Li
Optimal Location Queries in Road Network Databases
The OTF algorithm: construct the A(v ) on the fly
v6
c7
c9 c1 f1
c5
c6
c8 v1
f3 v5
v c2 2 c3
f2
v4 v3
c4
A(v2) < c2 , 1 > < c3 , 2 > < c8 , 3 >
Xiaokui Xiao, Bin Yao, Feifei Li
Optimal Location Queries in Road Network Databases
The OTF algorithm: construct the A(v ) on the fly
v6
c7
c9 c1 f1
c5
c6
c8 v1
f3 v5
v c2 2 c3
f2
v4 v3
c4
A(v2) < c2 , 1 > < c3 , 2 > < c8 , 3 >
Xiaokui Xiao, Bin Yao, Feifei Li
Optimal Location Queries in Road Network Databases
The OTF algorithm: construct the A(v ) on the fly
v6
c7
c9 c1 f1
c5
c6
c8 v1
f3 v5
v c2 2 c3
f2
v4 v3
c4
A(v2) < c2 , 1 > < c3 , 2 > < c8 , 3 > Time: O(n2 log n), space: O(n)
Xiaokui Xiao, Bin Yao, Feifei Li
Optimal Location Queries in Road Network Databases
Fine-grained Partitioning (FGP) Enumerating the local optima will incur significant overhead when Ec is large.
Xiaokui Xiao, Bin Yao, Feifei Li
Optimal Location Queries in Road Network Databases
Fine-grained Partitioning (FGP) Enumerating the local optima will incur significant overhead when Ec is large.
G3 v6 G1
c9 c1 f1
c7
c5
c6
c8 v1
f3 v5
v c2 2 c3
Xiaokui Xiao, Bin Yao, Feifei Li
f2
v4 v3
c4 G2
Optimal Location Queries in Road Network Databases
Experiment For each type of OL queries, we examine two approaches: the basic approach the Fine-grained partitioning (FGP) approach
Xiaokui Xiao, Bin Yao, Feifei Li
Optimal Location Queries in Road Network Databases
Experiment For each type of OL queries, we examine two approaches: the basic approach the Fine-grained partitioning (FGP) approach
For each approach, we test two techniques for deriving attraction sets: the Blossom and OTF.
Xiaokui Xiao, Bin Yao, Feifei Li
Optimal Location Queries in Road Network Databases
Experiment For each type of OL queries, we examine two approaches: the basic approach the Fine-grained partitioning (FGP) approach
For each approach, we test two techniques for deriving attraction sets: the Blossom and OTF. C++, Linux, Intel Xeon 2GHz CPU and 4GB memory
Xiaokui Xiao, Bin Yao, Feifei Li
Optimal Location Queries in Road Network Databases
Experiment For each type of OL queries, we examine two approaches: the basic approach the Fine-grained partitioning (FGP) approach
For each approach, we test two techniques for deriving attraction sets: the Blossom and OTF. C++, Linux, Intel Xeon 2GHz CPU and 4GB memory Data sets San Francisco(SF) and California(CA) road networks from the Digital Chart of the World Server. building locations in SF and CA from the OpenStreetMap project.
Xiaokui Xiao, Bin Yao, Feifei Li
Optimal Location Queries in Road Network Databases
Experiment For each type of OL queries, we examine two approaches: the basic approach the Fine-grained partitioning (FGP) approach
For each approach, we test two techniques for deriving attraction sets: the Blossom and OTF. C++, Linux, Intel Xeon 2GHz CPU and 4GB memory Data sets San Francisco(SF) and California(CA) road networks from the Digital Chart of the World Server. building locations in SF and CA from the OpenStreetMap project.
Default settings. Symbol |F | |C | τ
Definition number of facilities number of clients the percentage of candidate edges
Xiaokui Xiao, Bin Yao, Feifei Li
Default Value 1000 300, 000 100%
Optimal Location Queries in Road Network Databases
memory consumption (MB)
Vary |F | (CLQ on SF) Blossom
OTF Basic FGP
4
10
3
10
2
10
out of memory
1
10
0
10
0
0.5
1
2
3
3
4
number of facilities ×10 Xiaokui Xiao, Bin Yao, Feifei Li
Optimal Location Queries in Road Network Databases
Vary |F | (CLQ on SF) Blossom
running time (seconds)
5
OTF
10
Basic FGP
out of memory
4
10
3
10
2
10
1
10
0
0.5
1
2
3
3
4
number of facilities ×10 Xiaokui Xiao, Bin Yao, Feifei Li
Optimal Location Queries in Road Network Databases
memory consumption (MB)
Vary |C | (CLQ on SF) Blossom
4
10
3
OTF
Basic FGP
10
out of memory
2
10
1
10
0
10
0
1
2
3
4
5
5
number of clients ×10 Xiaokui Xiao, Bin Yao, Feifei Li
Optimal Location Queries in Road Network Databases
Vary |C | (CLQ on SF) Blossom
running time (seconds)
4
10
Basic FGP
OTF
out of memory
3
10
2
10
1
10
0
1
2
3
4
5
5
number of clients ×10 Xiaokui Xiao, Bin Yao, Feifei Li
Optimal Location Queries in Road Network Databases
Conclusion
Define three variants of OL queries on the road networks.
Xiaokui Xiao, Bin Yao, Feifei Li
Optimal Location Queries in Road Network Databases
Conclusion
Define three variants of OL queries on the road networks. Introduce a unified framework that addresses all three query types efficiently.
Xiaokui Xiao, Bin Yao, Feifei Li
Optimal Location Queries in Road Network Databases
Conclusion
Define three variants of OL queries on the road networks. Introduce a unified framework that addresses all three query types efficiently. Future work the incremental monitoring of the optimal locations when the facility or client sets have been updated. the optimal location queries for moving objects in road networks.
Xiaokui Xiao, Bin Yao, Feifei Li
Optimal Location Queries in Road Network Databases
The end
Thank You Q and A
Xiaokui Xiao, Bin Yao, Feifei Li
Optimal Location Queries in Road Network Databases
